![](\"https://miro.medium.com/max/1838/1*aOYUa3hHKi9HlYnnFAipUQ.gif\")
\n",
68 | "\n",
69 | "TensorFlow graph example (from [https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4](https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4)).\n",
70 | ":::\n",
71 | "\n",
72 | "\n",
73 | "#### Why do so many people like TensorFlow?\n",
74 | "TensorFlow is intentionally user-friendly, with helpful plugins to visualize model training and a useful software debugging tool. As well, TensorFlow is highly scalable, with easy deployment on both CPUs and GPUs."
75 | ]
76 | },
77 | {
78 | "cell_type": "markdown",
79 | "metadata": {},
80 | "source": [
81 | "### What is Keras?\n",
82 | "\n",
83 | "[Keras](https://keras.io/about/) is an API built on Python which reduces the cognitive load associated with programming models through human readability and simple and consistent structures.\n",
84 | "\n",
85 | "Keras is what some might call a wrapper for TensorFlow. It is intended for rapid experimentation.\n",
86 | "\n",
87 | "Tha main components of Keras include:\n",
88 | "1. A models API, which enables one to construct a model with varying levels of complexity depending on use case. We will use the [Functional API](https://keras.io/guides/functional_api/) subclass.\n",
89 | "2. A layers API, which allows one to define the tensor in/tensor out computation functions.\n",
90 | "3. A callback API, which enables one to program specific actions to occur during training, such as log training metrics, visualize interim/internal states and statistics of the model during training, and perform early stopping when the model converges.\n",
91 | "4. A data preprocessing API, which offers support for prepping raw data from disk to model ready Tensor format.\n",
92 | "5. An optimizer API where all of the state of the art optimizers can be plugged in. Learning rate decay / scheduling can also be implemented as part of this API.\n",
93 | "6. A metrics API which is used for assessing the performance of the model during training. A metric is the target to optimize during training, with specific metrics chosen for specific modeling objectives.\n",
94 | "7. A loss API that informs the model quantitatively how much it should try to minimize during training by providing a measure of error. Similar to metrics, specific loss functions are selected for specific modeling objectives.\n",
95 | "\n",
96 | "With the Functional API, our main workflow will follow the diagram below.\n",
97 | "\n",
98 | ":::{figure-md} Keras-fig\n",
99 | "![](\"images/Keras_functional_API.jpg\")
\n",
100 | "\n",
101 | "Keras Functional API diagram (from [https://miro.com/app/board/o9J_lhnKhVE=/](hhttps://miro.com/app/board/o9J_lhnKhVE=/)).\n",
102 | ":::"
103 | ]
104 | }
105 | ],
106 | "metadata": {
107 | "kernelspec": {
108 | "display_name": "Python 3",
109 | "language": "python",
110 | "name": "python3"
111 | },
112 | "language_info": {
113 | "codemirror_mode": {
114 | "name": "ipython",
115 | "version": 3
116 | },
117 | "file_extension": ".py",
118 | "mimetype": "text/x-python",
119 | "name": "python",
120 | "nbconvert_exporter": "python",
121 | "pygments_lexer": "ipython3",
122 | "version": "3.8.11"
123 | }
124 | },
125 | "nbformat": 4,
126 | "nbformat_minor": 2
127 | }
128 |
--------------------------------------------------------------------------------
/ds_book/_build/html/_sources/docs/Lesson1b_annex.ipynb:
--------------------------------------------------------------------------------
1 | {
2 | "cells": [
3 | {
4 | "cell_type": "markdown",
5 | "metadata": {},
6 | "source": [
7 | "# Introduction to TensorFlow and Keras"
8 | ]
9 | },
10 | {
11 | "cell_type": "markdown",
12 | "metadata": {},
13 | "source": [
14 | "## Objectives"
15 | ]
16 | },
17 | {
18 | "cell_type": "markdown",
19 | "metadata": {},
20 | "source": [
21 | "The goal of this notebook is to teach some basics of the TensorFlow framework and the Keras API."
22 | ]
23 | },
24 | {
25 | "cell_type": "markdown",
26 | "metadata": {},
27 | "source": [
28 | "### What is TensorFlow?\n",
29 | "\n",
30 | "[TensorFlow](https://www.tensorflow.org/guide]) is an open-source framework developed by Google for building various machine learning and deep learning models. Although originally released in late 2015, the first stable version arrived in 2017. TensorFlow is free and open-source, thanks to the Apache Open Source license.\n",
31 | "\n",
32 | "The main objective of using TensorFlow is to reduce the complexity of implementing computations on large numerical data sets. In practice, these large computations can manifest as training and inference with machine learning or deep learning models.\n",
33 | "\n",
34 | "TensorFlow was designed to operate with multiple CPUs or GPUs, as well as a growing number of mobile operating systems. The framework includes wrappers in Python, C++, and Java.\n",
35 | "\n",
36 | "#### How does it work?\n",
37 | "TensorFlow accepts inputs as a multi-dimensional array called a Tensor, which allows the programmer to create dataflow graphs and structures specifying how data travels through. The framework is designed to support creation of a flowchart of operations to be applied to input Tensors, which travel in one direction and out the other.\n",
38 | "\n",
39 | "#### TensorFlow’s structure\n",
40 | "There are three main components to TensorFlow's structure.\n",
41 | "\n",
42 | "1. preprocessing the data\n",
43 | "2. building the model\n",
44 | "3. training and estimating the model\n",
45 | "\n",
46 | "The name TensorFlow derives from the way in which the framework receives input in the form of a multi-dimensional array, i.e. the tensors. These tensors travel sequentially through the specified flowchart of the operations, entering at one end and culminate as output at the other end. \n",
47 | "\n",
48 | "#### What are TensorFlow components?\n",
49 | "**Tensor**\n",
50 | "\n",
51 | "Tensors, the basic unit of data in this framework, are involved in every computation of TensorFlow. A tensor is an n-dimensional vector or matrix. In theory, a tensor may represent any form of data. The values belonging to a tensor all share the same data type and often the same shape / dimensionality. A tensor can describe the input data and the output of a calculation. \n",
52 | "\n",
53 | "In TensorFlow, all operations are carried out within a graph, which in effect is a series of computations that happen in order. Each indivisual operation is referred to as an op node. \n",
54 | "\n",
55 | "**Graphs**\n",
56 | "\n",
57 | "TensorFlow uses a graph framework. Graphs collect and summarize all of the calculations and offer several benefits:\n",
58 | "\n",
59 | "1. They are designed to work on CPUs or GPUs, as well as on mobile devices.\n",
60 | "2. Graphs are portable which enables the computations to be saved for immediate or later usage. Otherwsie stated, the graph can be frozen and run at a later time.\n",
61 | "3. Graph calculations are executed by linking tensors together.\n",
62 | "4. For each tensor, there is a node and an edge. The node carries out the mathematical process and produces endpoint outputs. The input/output connections are represented by the edges.\n",
63 | "5. All nodes are linked together, so the graph itself is a depiction of the operations and relationships that exist between the nodes.\n",
64 | "\n",
65 | ":::{figure-md} TFgraph-fig\n",
66 | "\n",
67 | "\n",
68 | "TensorFlow graph example (from [https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4](https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4)).\n",
69 | ":::\n",
70 | "\n",
71 | "#### Why do so many people like TensorFlow?\n",
72 | "TensorFlow is intentionally user-friendly, with helpful plugins to visualize model training and a useful software debugging tool. As well, TensorFlow is highly scalable, with easy deployment on both CPUs and GPUs."
73 | ]
74 | },
75 | {
76 | "cell_type": "markdown",
77 | "metadata": {},
78 | "source": [
79 | "### What is Keras?\n",
80 | "\n",
81 | "[Keras](https://keras.io/about/) is an API built on Python, with human readability at the forefront of its design. With simple and consistent structures and methods extensible across many machine learning and deep learning applications, Keras reduces the cognitive load associated with programming models. Furthermore, the API seeks to minimize the need for programmer interaction by abstracting many complexities into easily callable functions. Lastly, Keras features clear & actionable error messaging, complemented by comprehensive and digestible documentation and developer guides.\n",
82 | "\n",
83 | "Keras is what some might call a wrapper for TensorFlow. By that, one means to say that Keras simplifies a programmer's interaction with TensorFlow through refinement of key methods and constructs.\n",
84 | "\n",
85 | "Importantly, Keras is intended for rapid experimentation.\n",
86 | "\n",
87 | "Tha main components of Keras include:\n",
88 | "1. A models API, which enables one to construct a model with varying levels of complexity depending on use case. We will use the [Functional API](https://keras.io/guides/functional_api/) subclass.\n",
89 | "2. A layers API, which allows one to define the tensor in/tnesor out computation functions.\n",
90 | "3. A callback API, which enables one to program specific actions to occur during training, such as log training metrics, visualize interim/internal states and statistics of the model during training, and perform early stopping when the model converges.\n",
91 | "4. A data preprocessing API, which offers support for prepping raw data from disk to model ready Tensor format.\n",
92 | "5. An optimizer API where all of the state of the art optimizers can be plugged in. Learning rate decay / scheduling can also be implemented a spart of this API.\n",
93 | "6. A metrics API which is used for assessing the performance of the model during training. A metric is the target to optimize during training, with specific metrics chosen for specific modeling objectives.\n",
94 | "7. A loss API that informs the model quantitatively how much it should try to minimize during training by providing a measure of error. Similar to metrics, specific loss functions are selected for specific modelung objectives.\n",
95 | "\n",
96 | "With the Functional API, our main workflow will follow the diagram below.\n",
97 | "\n",
98 | ":::{figure-md} Keras-fig\n",
99 | "\n",
100 | "\n",
101 | "Keras Functional API diagram (from [https://miro.com/app/board/o9J_lhnKhVE=/](hhttps://miro.com/app/board/o9J_lhnKhVE=/)).\n",
102 | ":::"
103 | ]
104 | }
105 | ],
106 | "metadata": {
107 | "kernelspec": {
108 | "display_name": "Python 3",
109 | "language": "python",
110 | "name": "python3"
111 | },
112 | "language_info": {
113 | "codemirror_mode": {
114 | "name": "ipython",
115 | "version": 3
116 | },
117 | "file_extension": ".py",
118 | "mimetype": "text/x-python",
119 | "name": "python",
120 | "nbconvert_exporter": "python",
121 | "pygments_lexer": "ipython3",
122 | "version": "3.8.11"
123 | }
124 | },
125 | "nbformat": 4,
126 | "nbformat_minor": 2
127 | }
128 |
--------------------------------------------------------------------------------
/ds_book/_build/html/_sources/docs/appendix.ipynb:
--------------------------------------------------------------------------------
1 | {
2 | "cells": [
3 | {
4 | "cell_type": "markdown",
5 | "metadata": {},
6 | "source": [
7 | "# Appendix"
8 | ]
9 | }
10 | ],
11 | "metadata": {
12 | "kernelspec": {
13 | "display_name": "Python 3",
14 | "language": "python",
15 | "name": "python3"
16 | },
17 | "language_info": {
18 | "codemirror_mode": {
19 | "name": "ipython",
20 | "version": 3
21 | },
22 | "file_extension": ".py",
23 | "mimetype": "text/x-python",
24 | "name": "python",
25 | "nbconvert_exporter": "python",
26 | "pygments_lexer": "ipython3",
27 | "version": "3.8.11"
28 | }
29 | },
30 | "nbformat": 4,
31 | "nbformat_minor": 4
32 | }
33 |
--------------------------------------------------------------------------------
/ds_book/_build/html/_sources/docs/index.md:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 | # Deep Learning with TensorFlow: Short
52 | */ 53 | .o-tooltip--left { 54 | position: relative; 55 | } 56 | 57 | .o-tooltip--left:after { 58 | opacity: 0; 59 | visibility: hidden; 60 | position: absolute; 61 | content: attr(data-tooltip); 62 | padding: .2em; 63 | font-size: .8em; 64 | left: -.2em; 65 | background: grey; 66 | color: white; 67 | white-space: nowrap; 68 | z-index: 2; 69 | border-radius: 2px; 70 | transform: translateX(-102%) translateY(0); 71 | transition: opacity 0.2s cubic-bezier(0.64, 0.09, 0.08, 1), transform 0.2s cubic-bezier(0.64, 0.09, 0.08, 1); 72 | } 73 | 74 | .o-tooltip--left:hover:after { 75 | display: block; 76 | opacity: 1; 77 | visibility: visible; 78 | transform: translateX(-100%) translateY(0); 79 | transition: opacity 0.2s cubic-bezier(0.64, 0.09, 0.08, 1), transform 0.2s cubic-bezier(0.64, 0.09, 0.08, 1); 80 | transition-delay: .5s; 81 | } 82 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/copybutton.js: -------------------------------------------------------------------------------- 1 | // Localization support 2 | const messages = { 3 | 'en': { 4 | 'copy': 'Copy', 5 | 'copy_to_clipboard': 'Copy to clipboard', 6 | 'copy_success': 'Copied!', 7 | 'copy_failure': 'Failed to copy', 8 | }, 9 | 'es' : { 10 | 'copy': 'Copiar', 11 | 'copy_to_clipboard': 'Copiar al portapapeles', 12 | 'copy_success': '¡Copiado!', 13 | 'copy_failure': 'Error al copiar', 14 | }, 15 | 'de' : { 16 | 'copy': 'Kopieren', 17 | 'copy_to_clipboard': 'In die Zwischenablage kopieren', 18 | 'copy_success': 'Kopiert!', 19 | 'copy_failure': 'Fehler beim Kopieren', 20 | }, 21 | 'fr' : { 22 | 'copy': 'Copier', 23 | 'copy_to_clipboard': 'Copié dans le presse-papier', 24 | 'copy_success': 'Copié !', 25 | 'copy_failure': 'Échec de la copie', 26 | }, 27 | 'ru': { 28 | 'copy': 'Скопировать', 29 | 'copy_to_clipboard': 'Скопировать в буфер', 30 | 'copy_success': 'Скопировано!', 31 | 'copy_failure': 'Не удалось скопировать', 32 | }, 33 | 'zh-CN': { 34 | 'copy': '复制', 35 | 'copy_to_clipboard': '复制到剪贴板', 36 | 'copy_success': '复制成功!', 37 | 'copy_failure': '复制失败', 38 | } 39 | } 40 | 41 | let locale = 'en' 42 | if( document.documentElement.lang !== undefined 43 | && messages[document.documentElement.lang] !== undefined ) { 44 | locale = document.documentElement.lang 45 | } 46 | 47 | let doc_url_root = DOCUMENTATION_OPTIONS.URL_ROOT; 48 | if (doc_url_root == '#') { 49 | doc_url_root = ''; 50 | } 51 | 52 | const path_static = `${doc_url_root}_static/`; 53 | 54 | /** 55 | * Set up copy/paste for code blocks 56 | */ 57 | 58 | const runWhenDOMLoaded = cb => { 59 | if (document.readyState != 'loading') { 60 | cb() 61 | } else if (document.addEventListener) { 62 | document.addEventListener('DOMContentLoaded', cb) 63 | } else { 64 | document.attachEvent('onreadystatechange', function() { 65 | if (document.readyState == 'complete') cb() 66 | }) 67 | } 68 | } 69 | 70 | const codeCellId = index => `codecell${index}` 71 | 72 | // Clears selected text since ClipboardJS will select the text when copying 73 | const clearSelection = () => { 74 | if (window.getSelection) { 75 | window.getSelection().removeAllRanges() 76 | } else if (document.selection) { 77 | document.selection.empty() 78 | } 79 | } 80 | 81 | // Changes tooltip text for two seconds, then changes it back 82 | const temporarilyChangeTooltip = (el, oldText, newText) => { 83 | el.setAttribute('data-tooltip', newText) 84 | el.classList.add('success') 85 | setTimeout(() => el.setAttribute('data-tooltip', oldText), 2000) 86 | setTimeout(() => el.classList.remove('success'), 2000) 87 | } 88 | 89 | // Changes the copy button icon for two seconds, then changes it back 90 | const temporarilyChangeIcon = (el) => { 91 | img = el.querySelector("img"); 92 | img.setAttribute('src', `${path_static}check-solid.svg`) 93 | setTimeout(() => img.setAttribute('src', `${path_static}copy-button.svg`), 2000) 94 | } 95 | 96 | const addCopyButtonToCodeCells = () => { 97 | // If ClipboardJS hasn't loaded, wait a bit and try again. This 98 | // happens because we load ClipboardJS asynchronously. 99 | if (window.ClipboardJS === undefined) { 100 | setTimeout(addCopyButtonToCodeCells, 250) 101 | return 102 | } 103 | 104 | // Add copybuttons to all of our code cells 105 | const codeCells = document.querySelectorAll('div.highlight pre') 106 | codeCells.forEach((codeCell, index) => { 107 | const id = codeCellId(index) 108 | codeCell.setAttribute('id', id) 109 | 110 | const clipboardButton = id => 111 | `` 114 | codeCell.insertAdjacentHTML('afterend', clipboardButton(id)) 115 | }) 116 | 117 | function escapeRegExp(string) { 118 | return string.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); // $& means the whole matched string 119 | } 120 | 121 | // Callback when a copy button is clicked. Will be passed the node that was clicked 122 | // should then grab the text and replace pieces of text that shouldn't be used in output 123 | function formatCopyText(textContent, copybuttonPromptText, isRegexp = false, onlyCopyPromptLines = true, removePrompts = true, copyEmptyLines = true, lineContinuationChar = "", hereDocDelim = "") { 124 | 125 | var regexp; 126 | var match; 127 | 128 | // Do we check for line continuation characters and "HERE-documents"? 129 | var useLineCont = !!lineContinuationChar 130 | var useHereDoc = !!hereDocDelim 131 | 132 | // create regexp to capture prompt and remaining line 133 | if (isRegexp) { 134 | regexp = new RegExp('^(' + copybuttonPromptText + ')(.*)') 135 | } else { 136 | regexp = new RegExp('^(' + escapeRegExp(copybuttonPromptText) + ')(.*)') 137 | } 138 | 139 | const outputLines = []; 140 | var promptFound = false; 141 | var gotLineCont = false; 142 | var gotHereDoc = false; 143 | const lineGotPrompt = []; 144 | for (const line of textContent.split('\n')) { 145 | match = line.match(regexp) 146 | if (match || gotLineCont || gotHereDoc) { 147 | promptFound = regexp.test(line) 148 | lineGotPrompt.push(promptFound) 149 | if (removePrompts && promptFound) { 150 | outputLines.push(match[2]) 151 | } else { 152 | outputLines.push(line) 153 | } 154 | gotLineCont = line.endsWith(lineContinuationChar) & useLineCont 155 | if (line.includes(hereDocDelim) & useHereDoc) 156 | gotHereDoc = !gotHereDoc 157 | } else if (!onlyCopyPromptLines) { 158 | outputLines.push(line) 159 | } else if (copyEmptyLines && line.trim() === '') { 160 | outputLines.push(line) 161 | } 162 | } 163 | 164 | // If no lines with the prompt were found then just use original lines 165 | if (lineGotPrompt.some(v => v === true)) { 166 | textContent = outputLines.join('\n'); 167 | } 168 | 169 | // Remove a trailing newline to avoid auto-running when pasting 170 | if (textContent.endsWith("\n")) { 171 | textContent = textContent.slice(0, -1) 172 | } 173 | return textContent 174 | } 175 | 176 | 177 | var copyTargetText = (trigger) => { 178 | var target = document.querySelector(trigger.attributes['data-clipboard-target'].value); 179 | return formatCopyText(target.innerText, '', false, true, true, true, '', '') 180 | } 181 | 182 | // Initialize with a callback so we can modify the text before copy 183 | const clipboard = new ClipboardJS('.copybtn', {text: copyTargetText}) 184 | 185 | // Update UI with error/success messages 186 | clipboard.on('success', event => { 187 | clearSelection() 188 | temporarilyChangeTooltip(event.trigger, messages[locale]['copy'], messages[locale]['copy_success']) 189 | temporarilyChangeIcon(event.trigger) 190 | }) 191 | 192 | clipboard.on('error', event => { 193 | temporarilyChangeTooltip(event.trigger, messages[locale]['copy'], messages[locale]['copy_failure']) 194 | }) 195 | } 196 | 197 | runWhenDOMLoaded(addCopyButtonToCodeCells) -------------------------------------------------------------------------------- /ds_book/_build/html/_static/copybutton_funcs.js: -------------------------------------------------------------------------------- 1 | function escapeRegExp(string) { 2 | return string.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); // $& means the whole matched string 3 | } 4 | 5 | // Callback when a copy button is clicked. Will be passed the node that was clicked 6 | // should then grab the text and replace pieces of text that shouldn't be used in output 7 | export function formatCopyText(textContent, copybuttonPromptText, isRegexp = false, onlyCopyPromptLines = true, removePrompts = true, copyEmptyLines = true, lineContinuationChar = "", hereDocDelim = "") { 8 | 9 | var regexp; 10 | var match; 11 | 12 | // Do we check for line continuation characters and "HERE-documents"? 13 | var useLineCont = !!lineContinuationChar 14 | var useHereDoc = !!hereDocDelim 15 | 16 | // create regexp to capture prompt and remaining line 17 | if (isRegexp) { 18 | regexp = new RegExp('^(' + copybuttonPromptText + ')(.*)') 19 | } else { 20 | regexp = new RegExp('^(' + escapeRegExp(copybuttonPromptText) + ')(.*)') 21 | } 22 | 23 | const outputLines = []; 24 | var promptFound = false; 25 | var gotLineCont = false; 26 | var gotHereDoc = false; 27 | const lineGotPrompt = []; 28 | for (const line of textContent.split('\n')) { 29 | match = line.match(regexp) 30 | if (match || gotLineCont || gotHereDoc) { 31 | promptFound = regexp.test(line) 32 | lineGotPrompt.push(promptFound) 33 | if (removePrompts && promptFound) { 34 | outputLines.push(match[2]) 35 | } else { 36 | outputLines.push(line) 37 | } 38 | gotLineCont = line.endsWith(lineContinuationChar) & useLineCont 39 | if (line.includes(hereDocDelim) & useHereDoc) 40 | gotHereDoc = !gotHereDoc 41 | } else if (!onlyCopyPromptLines) { 42 | outputLines.push(line) 43 | } else if (copyEmptyLines && line.trim() === '') { 44 | outputLines.push(line) 45 | } 46 | } 47 | 48 | // If no lines with the prompt were found then just use original lines 49 | if (lineGotPrompt.some(v => v === true)) { 50 | textContent = outputLines.join('\n'); 51 | } 52 | 53 | // Remove a trailing newline to avoid auto-running when pasting 54 | if (textContent.endsWith("\n")) { 55 | textContent = textContent.slice(0, -1) 56 | } 57 | return textContent 58 | } 59 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/css/theme.css: -------------------------------------------------------------------------------- 1 | :root { 2 | /***************************************************************************** 3 | * Theme config 4 | **/ 5 | --pst-header-height: 60px; 6 | 7 | /***************************************************************************** 8 | * Font size 9 | **/ 10 | --pst-font-size-base: 15px; /* base font size - applied at body / html level */ 11 | 12 | /* heading font sizes */ 13 | --pst-font-size-h1: 36px; 14 | --pst-font-size-h2: 32px; 15 | --pst-font-size-h3: 26px; 16 | --pst-font-size-h4: 21px; 17 | --pst-font-size-h5: 18px; 18 | --pst-font-size-h6: 16px; 19 | 20 | /* smaller then heading font sizes*/ 21 | --pst-font-size-milli: 12px; 22 | 23 | --pst-sidebar-font-size: .9em; 24 | --pst-sidebar-caption-font-size: .9em; 25 | 26 | /***************************************************************************** 27 | * Font family 28 | **/ 29 | /* These are adapted from https://systemfontstack.com/ */ 30 | --pst-font-family-base-system: -apple-system, BlinkMacSystemFont, Segoe UI, "Helvetica Neue", 31 | Arial, sans-serif, Apple Color Emoji, Segoe UI Emoji, Segoe UI Symbol; 32 | --pst-font-family-monospace-system: "SFMono-Regular", Menlo, Consolas, Monaco, 33 | Liberation Mono, Lucida Console, monospace; 34 | 35 | --pst-font-family-base: var(--pst-font-family-base-system); 36 | --pst-font-family-heading: var(--pst-font-family-base); 37 | --pst-font-family-monospace: var(--pst-font-family-monospace-system); 38 | 39 | /***************************************************************************** 40 | * Color 41 | * 42 | * Colors are defined in rgb string way, "red, green, blue" 43 | **/ 44 | --pst-color-primary: 19, 6, 84; 45 | --pst-color-success: 40, 167, 69; 46 | --pst-color-info: 0, 123, 255; /*23, 162, 184;*/ 47 | --pst-color-warning: 255, 193, 7; 48 | --pst-color-danger: 220, 53, 69; 49 | --pst-color-text-base: 51, 51, 51; 50 | 51 | --pst-color-h1: var(--pst-color-primary); 52 | --pst-color-h2: var(--pst-color-primary); 53 | --pst-color-h3: var(--pst-color-text-base); 54 | --pst-color-h4: var(--pst-color-text-base); 55 | --pst-color-h5: var(--pst-color-text-base); 56 | --pst-color-h6: var(--pst-color-text-base); 57 | --pst-color-paragraph: var(--pst-color-text-base); 58 | --pst-color-link: 0, 91, 129; 59 | --pst-color-link-hover: 227, 46, 0; 60 | --pst-color-headerlink: 198, 15, 15; 61 | --pst-color-headerlink-hover: 255, 255, 255; 62 | --pst-color-preformatted-text: 34, 34, 34; 63 | --pst-color-preformatted-background: 250, 250, 250; 64 | --pst-color-inline-code: 232, 62, 140; 65 | 66 | --pst-color-active-navigation: 19, 6, 84; 67 | --pst-color-navbar-link: 77, 77, 77; 68 | --pst-color-navbar-link-hover: var(--pst-color-active-navigation); 69 | --pst-color-navbar-link-active: var(--pst-color-active-navigation); 70 | --pst-color-sidebar-link: 77, 77, 77; 71 | --pst-color-sidebar-link-hover: var(--pst-color-active-navigation); 72 | --pst-color-sidebar-link-active: var(--pst-color-active-navigation); 73 | --pst-color-sidebar-expander-background-hover: 244, 244, 244; 74 | --pst-color-sidebar-caption: 77, 77, 77; 75 | --pst-color-toc-link: 119, 117, 122; 76 | --pst-color-toc-link-hover: var(--pst-color-active-navigation); 77 | --pst-color-toc-link-active: var(--pst-color-active-navigation); 78 | 79 | /***************************************************************************** 80 | * Icon 81 | **/ 82 | 83 | /* font awesome icons*/ 84 | --pst-icon-check-circle: '\f058'; 85 | --pst-icon-info-circle: '\f05a'; 86 | --pst-icon-exclamation-triangle: '\f071'; 87 | --pst-icon-exclamation-circle: '\f06a'; 88 | --pst-icon-times-circle: '\f057'; 89 | --pst-icon-lightbulb: '\f0eb'; 90 | 91 | /***************************************************************************** 92 | * Admonitions 93 | **/ 94 | 95 | --pst-color-admonition-default: var(--pst-color-info); 96 | --pst-color-admonition-note: var(--pst-color-info); 97 | --pst-color-admonition-attention: var(--pst-color-warning); 98 | --pst-color-admonition-caution: var(--pst-color-warning); 99 | --pst-color-admonition-warning: var(--pst-color-warning); 100 | --pst-color-admonition-danger: var(--pst-color-danger); 101 | --pst-color-admonition-error: var(--pst-color-danger); 102 | --pst-color-admonition-hint: var(--pst-color-success); 103 | --pst-color-admonition-tip: var(--pst-color-success); 104 | --pst-color-admonition-important: var(--pst-color-success); 105 | 106 | --pst-icon-admonition-default: var(--pst-icon-info-circle); 107 | --pst-icon-admonition-note: var(--pst-icon-info-circle); 108 | --pst-icon-admonition-attention: var(--pst-icon-exclamation-circle); 109 | --pst-icon-admonition-caution: var(--pst-icon-exclamation-triangle); 110 | --pst-icon-admonition-warning: var(--pst-icon-exclamation-triangle); 111 | --pst-icon-admonition-danger: var(--pst-icon-exclamation-triangle); 112 | --pst-icon-admonition-error: var(--pst-icon-times-circle); 113 | --pst-icon-admonition-hint: var(--pst-icon-lightbulb); 114 | --pst-icon-admonition-tip: var(--pst-icon-lightbulb); 115 | --pst-icon-admonition-important: var(--pst-icon-exclamation-circle); 116 | 117 | } 118 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/doctools.js: -------------------------------------------------------------------------------- 1 | /* 2 | * doctools.js 3 | * ~~~~~~~~~~~ 4 | * 5 | * Sphinx JavaScript utilities for all documentation. 6 | * 7 | * :copyright: Copyright 2007-2021 by the Sphinx team, see AUTHORS. 8 | * :license: BSD, see LICENSE for details. 9 | * 10 | */ 11 | 12 | /** 13 | * select a different prefix for underscore 14 | */ 15 | $u = _.noConflict(); 16 | 17 | /** 18 | * make the code below compatible with browsers without 19 | * an installed firebug like debugger 20 | if (!window.console || !console.firebug) { 21 | var names = ["log", "debug", "info", "warn", "error", "assert", "dir", 22 | "dirxml", "group", "groupEnd", "time", "timeEnd", "count", "trace", 23 | "profile", "profileEnd"]; 24 | window.console = {}; 25 | for (var i = 0; i < names.length; ++i) 26 | window.console[names[i]] = function() {}; 27 | } 28 | */ 29 | 30 | /** 31 | * small helper function to urldecode strings 32 | * 33 | * See https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/decodeURIComponent#Decoding_query_parameters_from_a_URL 34 | */ 35 | jQuery.urldecode = function(x) { 36 | if (!x) { 37 | return x 38 | } 39 | return decodeURIComponent(x.replace(/\+/g, ' ')); 40 | }; 41 | 42 | /** 43 | * small helper function to urlencode strings 44 | */ 45 | jQuery.urlencode = encodeURIComponent; 46 | 47 | /** 48 | * This function returns the parsed url parameters of the 49 | * current request. Multiple values per key are supported, 50 | * it will always return arrays of strings for the value parts. 51 | */ 52 | jQuery.getQueryParameters = function(s) { 53 | if (typeof s === 'undefined') 54 | s = document.location.search; 55 | var parts = s.substr(s.indexOf('?') + 1).split('&'); 56 | var result = {}; 57 | for (var i = 0; i < parts.length; i++) { 58 | var tmp = parts[i].split('=', 2); 59 | var key = jQuery.urldecode(tmp[0]); 60 | var value = jQuery.urldecode(tmp[1]); 61 | if (key in result) 62 | result[key].push(value); 63 | else 64 | result[key] = [value]; 65 | } 66 | return result; 67 | }; 68 | 69 | /** 70 | * highlight a given string on a jquery object by wrapping it in 71 | * span elements with the given class name. 72 | */ 73 | jQuery.fn.highlightText = function(text, className) { 74 | function highlight(node, addItems) { 75 | if (node.nodeType === 3) { 76 | var val = node.nodeValue; 77 | var pos = val.toLowerCase().indexOf(text); 78 | if (pos >= 0 && 79 | !jQuery(node.parentNode).hasClass(className) && 80 | !jQuery(node.parentNode).hasClass("nohighlight")) { 81 | var span; 82 | var isInSVG = jQuery(node).closest("body, svg, foreignObject").is("svg"); 83 | if (isInSVG) { 84 | span = document.createElementNS("http://www.w3.org/2000/svg", "tspan"); 85 | } else { 86 | span = document.createElement("span"); 87 | span.className = className; 88 | } 89 | span.appendChild(document.createTextNode(val.substr(pos, text.length))); 90 | node.parentNode.insertBefore(span, node.parentNode.insertBefore( 91 | document.createTextNode(val.substr(pos + text.length)), 92 | node.nextSibling)); 93 | node.nodeValue = val.substr(0, pos); 94 | if (isInSVG) { 95 | var rect = document.createElementNS("http://www.w3.org/2000/svg", "rect"); 96 | var bbox = node.parentElement.getBBox(); 97 | rect.x.baseVal.value = bbox.x; 98 | rect.y.baseVal.value = bbox.y; 99 | rect.width.baseVal.value = bbox.width; 100 | rect.height.baseVal.value = bbox.height; 101 | rect.setAttribute('class', className); 102 | addItems.push({ 103 | "parent": node.parentNode, 104 | "target": rect}); 105 | } 106 | } 107 | } 108 | else if (!jQuery(node).is("button, select, textarea")) { 109 | jQuery.each(node.childNodes, function() { 110 | highlight(this, addItems); 111 | }); 112 | } 113 | } 114 | var addItems = []; 115 | var result = this.each(function() { 116 | highlight(this, addItems); 117 | }); 118 | for (var i = 0; i < addItems.length; ++i) { 119 | jQuery(addItems[i].parent).before(addItems[i].target); 120 | } 121 | return result; 122 | }; 123 | 124 | /* 125 | * backward compatibility for jQuery.browser 126 | * This will be supported until firefox bug is fixed. 127 | */ 128 | if (!jQuery.browser) { 129 | jQuery.uaMatch = function(ua) { 130 | ua = ua.toLowerCase(); 131 | 132 | var match = /(chrome)[ \/]([\w.]+)/.exec(ua) || 133 | /(webkit)[ \/]([\w.]+)/.exec(ua) || 134 | /(opera)(?:.*version|)[ \/]([\w.]+)/.exec(ua) || 135 | /(msie) ([\w.]+)/.exec(ua) || 136 | ua.indexOf("compatible") < 0 && /(mozilla)(?:.*? rv:([\w.]+)|)/.exec(ua) || 137 | []; 138 | 139 | return { 140 | browser: match[ 1 ] || "", 141 | version: match[ 2 ] || "0" 142 | }; 143 | }; 144 | jQuery.browser = {}; 145 | jQuery.browser[jQuery.uaMatch(navigator.userAgent).browser] = true; 146 | } 147 | 148 | /** 149 | * Small JavaScript module for the documentation. 150 | */ 151 | var Documentation = { 152 | 153 | init : function() { 154 | this.fixFirefoxAnchorBug(); 155 | this.highlightSearchWords(); 156 | this.initIndexTable(); 157 | if (DOCUMENTATION_OPTIONS.NAVIGATION_WITH_KEYS) { 158 | this.initOnKeyListeners(); 159 | } 160 | }, 161 | 162 | /** 163 | * i18n support 164 | */ 165 | TRANSLATIONS : {}, 166 | PLURAL_EXPR : function(n) { return n === 1 ? 0 : 1; }, 167 | LOCALE : 'unknown', 168 | 169 | // gettext and ngettext don't access this so that the functions 170 | // can safely bound to a different name (_ = Documentation.gettext) 171 | gettext : function(string) { 172 | var translated = Documentation.TRANSLATIONS[string]; 173 | if (typeof translated === 'undefined') 174 | return string; 175 | return (typeof translated === 'string') ? translated : translated[0]; 176 | }, 177 | 178 | ngettext : function(singular, plural, n) { 179 | var translated = Documentation.TRANSLATIONS[singular]; 180 | if (typeof translated === 'undefined') 181 | return (n == 1) ? singular : plural; 182 | return translated[Documentation.PLURALEXPR(n)]; 183 | }, 184 | 185 | addTranslations : function(catalog) { 186 | for (var key in catalog.messages) 187 | this.TRANSLATIONS[key] = catalog.messages[key]; 188 | this.PLURAL_EXPR = new Function('n', 'return +(' + catalog.plural_expr + ')'); 189 | this.LOCALE = catalog.locale; 190 | }, 191 | 192 | /** 193 | * add context elements like header anchor links 194 | */ 195 | addContextElements : function() { 196 | $('div[id] > :header:first').each(function() { 197 | $('\u00B6'). 198 | attr('href', '#' + this.id). 199 | attr('title', _('Permalink to this headline')). 200 | appendTo(this); 201 | }); 202 | $('dt[id]').each(function() { 203 | $('\u00B6'). 204 | attr('href', '#' + this.id). 205 | attr('title', _('Permalink to this definition')). 206 | appendTo(this); 207 | }); 208 | }, 209 | 210 | /** 211 | * workaround a firefox stupidity 212 | * see: https://bugzilla.mozilla.org/show_bug.cgi?id=645075 213 | */ 214 | fixFirefoxAnchorBug : function() { 215 | if (document.location.hash && $.browser.mozilla) 216 | window.setTimeout(function() { 217 | document.location.href += ''; 218 | }, 10); 219 | }, 220 | 221 | /** 222 | * highlight the search words provided in the url in the text 223 | */ 224 | highlightSearchWords : function() { 225 | var params = $.getQueryParameters(); 226 | var terms = (params.highlight) ? params.highlight[0].split(/\s+/) : []; 227 | if (terms.length) { 228 | var body = $('div.body'); 229 | if (!body.length) { 230 | body = $('body'); 231 | } 232 | window.setTimeout(function() { 233 | $.each(terms, function() { 234 | body.highlightText(this.toLowerCase(), 'highlighted'); 235 | }); 236 | }, 10); 237 | $('' + _('Hide Search Matches') + '
') 239 | .appendTo($('#searchbox')); 240 | } 241 | }, 242 | 243 | /** 244 | * init the domain index toggle buttons 245 | */ 246 | initIndexTable : function() { 247 | var togglers = $('img.toggler').click(function() { 248 | var src = $(this).attr('src'); 249 | var idnum = $(this).attr('id').substr(7); 250 | $('tr.cg-' + idnum).toggle(); 251 | if (src.substr(-9) === 'minus.png') 252 | $(this).attr('src', src.substr(0, src.length-9) + 'plus.png'); 253 | else 254 | $(this).attr('src', src.substr(0, src.length-8) + 'minus.png'); 255 | }).css('display', ''); 256 | if (DOCUMENTATION_OPTIONS.COLLAPSE_INDEX) { 257 | togglers.click(); 258 | } 259 | }, 260 | 261 | /** 262 | * helper function to hide the search marks again 263 | */ 264 | hideSearchWords : function() { 265 | $('#searchbox .highlight-link').fadeOut(300); 266 | $('span.highlighted').removeClass('highlighted'); 267 | }, 268 | 269 | /** 270 | * make the url absolute 271 | */ 272 | makeURL : function(relativeURL) { 273 | return DOCUMENTATION_OPTIONS.URL_ROOT + '/' + relativeURL; 274 | }, 275 | 276 | /** 277 | * get the current relative url 278 | */ 279 | getCurrentURL : function() { 280 | var path = document.location.pathname; 281 | var parts = path.split(/\//); 282 | $.each(DOCUMENTATION_OPTIONS.URL_ROOT.split(/\//), function() { 283 | if (this === '..') 284 | parts.pop(); 285 | }); 286 | var url = parts.join('/'); 287 | return path.substring(url.lastIndexOf('/') + 1, path.length - 1); 288 | }, 289 | 290 | initOnKeyListeners: function() { 291 | $(document).keydown(function(event) { 292 | var activeElementType = document.activeElement.tagName; 293 | // don't navigate when in search box, textarea, dropdown or button 294 | if (activeElementType !== 'TEXTAREA' && activeElementType !== 'INPUT' && activeElementType !== 'SELECT' 295 | && activeElementType !== 'BUTTON' && !event.altKey && !event.ctrlKey && !event.metaKey 296 | && !event.shiftKey) { 297 | switch (event.keyCode) { 298 | case 37: // left 299 | var prevHref = $('link[rel="prev"]').prop('href'); 300 | if (prevHref) { 301 | window.location.href = prevHref; 302 | return false; 303 | } 304 | case 39: // right 305 | var nextHref = $('link[rel="next"]').prop('href'); 306 | if (nextHref) { 307 | window.location.href = nextHref; 308 | return false; 309 | } 310 | } 311 | } 312 | }); 313 | } 314 | }; 315 | 316 | // quick alias for translations 317 | _ = Documentation.gettext; 318 | 319 | $(document).ready(function() { 320 | Documentation.init(); 321 | }); 322 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/documentation_options.js: -------------------------------------------------------------------------------- 1 | var DOCUMENTATION_OPTIONS = { 2 | URL_ROOT: document.getElementById("documentation_options").getAttribute('data-url_root'), 3 | VERSION: '', 4 | LANGUAGE: 'None', 5 | COLLAPSE_INDEX: false, 6 | BUILDER: 'html', 7 | FILE_SUFFIX: '.html', 8 | LINK_SUFFIX: '.html', 9 | HAS_SOURCE: true, 10 | SOURCELINK_SUFFIX: '', 11 | NAVIGATION_WITH_KEYS: true 12 | }; -------------------------------------------------------------------------------- /ds_book/_build/html/_static/ds.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/html/_static/ds.png -------------------------------------------------------------------------------- /ds_book/_build/html/_static/file.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/html/_static/file.png -------------------------------------------------------------------------------- /ds_book/_build/html/_static/images/logo_binder.svg: -------------------------------------------------------------------------------- 1 | 2 | 3 | 20 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/images/logo_colab.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/html/_static/images/logo_colab.png -------------------------------------------------------------------------------- /ds_book/_build/html/_static/images/logo_jupyterhub.svg: -------------------------------------------------------------------------------- 1 | 2 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/language_data.js: -------------------------------------------------------------------------------- 1 | /* 2 | * language_data.js 3 | * ~~~~~~~~~~~~~~~~ 4 | * 5 | * This script contains the language-specific data used by searchtools.js, 6 | * namely the list of stopwords, stemmer, scorer and splitter. 7 | * 8 | * :copyright: Copyright 2007-2021 by the Sphinx team, see AUTHORS. 9 | * :license: BSD, see LICENSE for details. 10 | * 11 | */ 12 | 13 | var stopwords = ["a","and","are","as","at","be","but","by","for","if","in","into","is","it","near","no","not","of","on","or","such","that","the","their","then","there","these","they","this","to","was","will","with"]; 14 | 15 | 16 | /* Non-minified version is copied as a separate JS file, is available */ 17 | 18 | /** 19 | * Porter Stemmer 20 | */ 21 | var Stemmer = function() { 22 | 23 | var step2list = { 24 | ational: 'ate', 25 | tional: 'tion', 26 | enci: 'ence', 27 | anci: 'ance', 28 | izer: 'ize', 29 | bli: 'ble', 30 | alli: 'al', 31 | entli: 'ent', 32 | eli: 'e', 33 | ousli: 'ous', 34 | ization: 'ize', 35 | ation: 'ate', 36 | ator: 'ate', 37 | alism: 'al', 38 | iveness: 'ive', 39 | fulness: 'ful', 40 | ousness: 'ous', 41 | aliti: 'al', 42 | iviti: 'ive', 43 | biliti: 'ble', 44 | logi: 'log' 45 | }; 46 | 47 | var step3list = { 48 | icate: 'ic', 49 | ative: '', 50 | alize: 'al', 51 | iciti: 'ic', 52 | ical: 'ic', 53 | ful: '', 54 | ness: '' 55 | }; 56 | 57 | var c = "[^aeiou]"; // consonant 58 | var v = "[aeiouy]"; // vowel 59 | var C = c + "[^aeiouy]*"; // consonant sequence 60 | var V = v + "[aeiou]*"; // vowel sequence 61 | 62 | var mgr0 = "^(" + C + ")?" + V + C; // [C]VC... is m>0 63 | var meq1 = "^(" + C + ")?" + V + C + "(" + V + ")?$"; // [C]VC[V] is m=1 64 | var mgr1 = "^(" + C + ")?" + V + C + V + C; // [C]VCVC... is m>1 65 | var s_v = "^(" + C + ")?" + v; // vowel in stem 66 | 67 | this.stemWord = function (w) { 68 | var stem; 69 | var suffix; 70 | var firstch; 71 | var origword = w; 72 | 73 | if (w.length < 3) 74 | return w; 75 | 76 | var re; 77 | var re2; 78 | var re3; 79 | var re4; 80 | 81 | firstch = w.substr(0,1); 82 | if (firstch == "y") 83 | w = firstch.toUpperCase() + w.substr(1); 84 | 85 | // Step 1a 86 | re = /^(.+?)(ss|i)es$/; 87 | re2 = /^(.+?)([^s])s$/; 88 | 89 | if (re.test(w)) 90 | w = w.replace(re,"$1$2"); 91 | else if (re2.test(w)) 92 | w = w.replace(re2,"$1$2"); 93 | 94 | // Step 1b 95 | re = /^(.+?)eed$/; 96 | re2 = /^(.+?)(ed|ing)$/; 97 | if (re.test(w)) { 98 | var fp = re.exec(w); 99 | re = new RegExp(mgr0); 100 | if (re.test(fp[1])) { 101 | re = /.$/; 102 | w = w.replace(re,""); 103 | } 104 | } 105 | else if (re2.test(w)) { 106 | var fp = re2.exec(w); 107 | stem = fp[1]; 108 | re2 = new RegExp(s_v); 109 | if (re2.test(stem)) { 110 | w = stem; 111 | re2 = /(at|bl|iz)$/; 112 | re3 = new RegExp("([^aeiouylsz])\\1$"); 113 | re4 = new RegExp("^" + C + v + "[^aeiouwxy]$"); 114 | if (re2.test(w)) 115 | w = w + "e"; 116 | else if (re3.test(w)) { 117 | re = /.$/; 118 | w = w.replace(re,""); 119 | } 120 | else if (re4.test(w)) 121 | w = w + "e"; 122 | } 123 | } 124 | 125 | // Step 1c 126 | re = /^(.+?)y$/; 127 | if (re.test(w)) { 128 | var fp = re.exec(w); 129 | stem = fp[1]; 130 | re = new RegExp(s_v); 131 | if (re.test(stem)) 132 | w = stem + "i"; 133 | } 134 | 135 | // Step 2 136 | re = /^(.+?)(ational|tional|enci|anci|izer|bli|alli|entli|eli|ousli|ization|ation|ator|alism|iveness|fulness|ousness|aliti|iviti|biliti|logi)$/; 137 | if (re.test(w)) { 138 | var fp = re.exec(w); 139 | stem = fp[1]; 140 | suffix = fp[2]; 141 | re = new RegExp(mgr0); 142 | if (re.test(stem)) 143 | w = stem + step2list[suffix]; 144 | } 145 | 146 | // Step 3 147 | re = /^(.+?)(icate|ative|alize|iciti|ical|ful|ness)$/; 148 | if (re.test(w)) { 149 | var fp = re.exec(w); 150 | stem = fp[1]; 151 | suffix = fp[2]; 152 | re = new RegExp(mgr0); 153 | if (re.test(stem)) 154 | w = stem + step3list[suffix]; 155 | } 156 | 157 | // Step 4 158 | re = /^(.+?)(al|ance|ence|er|ic|able|ible|ant|ement|ment|ent|ou|ism|ate|iti|ous|ive|ize)$/; 159 | re2 = /^(.+?)(s|t)(ion)$/; 160 | if (re.test(w)) { 161 | var fp = re.exec(w); 162 | stem = fp[1]; 163 | re = new RegExp(mgr1); 164 | if (re.test(stem)) 165 | w = stem; 166 | } 167 | else if (re2.test(w)) { 168 | var fp = re2.exec(w); 169 | stem = fp[1] + fp[2]; 170 | re2 = new RegExp(mgr1); 171 | if (re2.test(stem)) 172 | w = stem; 173 | } 174 | 175 | // Step 5 176 | re = /^(.+?)e$/; 177 | if (re.test(w)) { 178 | var fp = re.exec(w); 179 | stem = fp[1]; 180 | re = new RegExp(mgr1); 181 | re2 = new RegExp(meq1); 182 | re3 = new RegExp("^" + C + v + "[^aeiouwxy]$"); 183 | if (re.test(stem) || (re2.test(stem) && !(re3.test(stem)))) 184 | w = stem; 185 | } 186 | re = /ll$/; 187 | re2 = new RegExp(mgr1); 188 | if (re.test(w) && re2.test(w)) { 189 | re = /.$/; 190 | w = w.replace(re,""); 191 | } 192 | 193 | // and turn initial Y back to y 194 | if (firstch == "y") 195 | w = firstch.toLowerCase() + w.substr(1); 196 | return w; 197 | } 198 | } 199 | 200 | 201 | 202 | 203 | var splitChars = (function() { 204 | var result = {}; 205 | var singles = [96, 180, 187, 191, 215, 247, 749, 885, 903, 907, 909, 930, 1014, 1648, 206 | 1748, 1809, 2416, 2473, 2481, 2526, 2601, 2609, 2612, 2615, 2653, 2702, 207 | 2706, 2729, 2737, 2740, 2857, 2865, 2868, 2910, 2928, 2948, 2961, 2971, 208 | 2973, 3085, 3089, 3113, 3124, 3213, 3217, 3241, 3252, 3295, 3341, 3345, 209 | 3369, 3506, 3516, 3633, 3715, 3721, 3736, 3744, 3748, 3750, 3756, 3761, 210 | 3781, 3912, 4239, 4347, 4681, 4695, 4697, 4745, 4785, 4799, 4801, 4823, 211 | 4881, 5760, 5901, 5997, 6313, 7405, 8024, 8026, 8028, 8030, 8117, 8125, 212 | 8133, 8181, 8468, 8485, 8487, 8489, 8494, 8527, 11311, 11359, 11687, 11695, 213 | 11703, 11711, 11719, 11727, 11735, 12448, 12539, 43010, 43014, 43019, 43587, 214 | 43696, 43713, 64286, 64297, 64311, 64317, 64319, 64322, 64325, 65141]; 215 | var i, j, start, end; 216 | for (i = 0; i < singles.length; i++) { 217 | result[singles[i]] = true; 218 | } 219 | var ranges = [[0, 47], [58, 64], [91, 94], [123, 169], [171, 177], [182, 184], [706, 709], 220 | [722, 735], [741, 747], [751, 879], [888, 889], [894, 901], [1154, 1161], 221 | [1318, 1328], [1367, 1368], [1370, 1376], [1416, 1487], [1515, 1519], [1523, 1568], 222 | [1611, 1631], [1642, 1645], [1750, 1764], [1767, 1773], [1789, 1790], [1792, 1807], 223 | [1840, 1868], [1958, 1968], [1970, 1983], [2027, 2035], [2038, 2041], [2043, 2047], 224 | [2070, 2073], [2075, 2083], [2085, 2087], [2089, 2307], [2362, 2364], [2366, 2383], 225 | [2385, 2391], [2402, 2405], [2419, 2424], [2432, 2436], [2445, 2446], [2449, 2450], 226 | [2483, 2485], [2490, 2492], [2494, 2509], [2511, 2523], [2530, 2533], [2546, 2547], 227 | [2554, 2564], [2571, 2574], [2577, 2578], [2618, 2648], [2655, 2661], [2672, 2673], 228 | [2677, 2692], [2746, 2748], [2750, 2767], [2769, 2783], [2786, 2789], [2800, 2820], 229 | [2829, 2830], [2833, 2834], [2874, 2876], [2878, 2907], [2914, 2917], [2930, 2946], 230 | [2955, 2957], [2966, 2968], [2976, 2978], [2981, 2983], [2987, 2989], [3002, 3023], 231 | [3025, 3045], [3059, 3076], [3130, 3132], [3134, 3159], [3162, 3167], [3170, 3173], 232 | [3184, 3191], [3199, 3204], [3258, 3260], [3262, 3293], [3298, 3301], [3312, 3332], 233 | [3386, 3388], [3390, 3423], [3426, 3429], [3446, 3449], [3456, 3460], [3479, 3481], 234 | [3518, 3519], [3527, 3584], [3636, 3647], [3655, 3663], [3674, 3712], [3717, 3718], 235 | [3723, 3724], [3726, 3731], [3752, 3753], [3764, 3772], [3774, 3775], [3783, 3791], 236 | [3802, 3803], [3806, 3839], [3841, 3871], [3892, 3903], [3949, 3975], [3980, 4095], 237 | [4139, 4158], [4170, 4175], [4182, 4185], [4190, 4192], [4194, 4196], [4199, 4205], 238 | [4209, 4212], [4226, 4237], [4250, 4255], [4294, 4303], [4349, 4351], [4686, 4687], 239 | [4702, 4703], [4750, 4751], [4790, 4791], [4806, 4807], [4886, 4887], [4955, 4968], 240 | [4989, 4991], [5008, 5023], [5109, 5120], [5741, 5742], [5787, 5791], [5867, 5869], 241 | [5873, 5887], [5906, 5919], [5938, 5951], [5970, 5983], [6001, 6015], [6068, 6102], 242 | [6104, 6107], [6109, 6111], [6122, 6127], [6138, 6159], [6170, 6175], [6264, 6271], 243 | [6315, 6319], [6390, 6399], [6429, 6469], [6510, 6511], [6517, 6527], [6572, 6592], 244 | [6600, 6607], [6619, 6655], [6679, 6687], [6741, 6783], [6794, 6799], [6810, 6822], 245 | [6824, 6916], [6964, 6980], [6988, 6991], [7002, 7042], [7073, 7085], [7098, 7167], 246 | [7204, 7231], [7242, 7244], [7294, 7400], [7410, 7423], [7616, 7679], [7958, 7959], 247 | [7966, 7967], [8006, 8007], [8014, 8015], [8062, 8063], [8127, 8129], [8141, 8143], 248 | [8148, 8149], [8156, 8159], [8173, 8177], [8189, 8303], [8306, 8307], [8314, 8318], 249 | [8330, 8335], [8341, 8449], [8451, 8454], [8456, 8457], [8470, 8472], [8478, 8483], 250 | [8506, 8507], [8512, 8516], [8522, 8525], [8586, 9311], [9372, 9449], [9472, 10101], 251 | [10132, 11263], [11493, 11498], [11503, 11516], [11518, 11519], [11558, 11567], 252 | [11622, 11630], [11632, 11647], [11671, 11679], [11743, 11822], [11824, 12292], 253 | [12296, 12320], [12330, 12336], [12342, 12343], [12349, 12352], [12439, 12444], 254 | [12544, 12548], [12590, 12592], [12687, 12689], [12694, 12703], [12728, 12783], 255 | [12800, 12831], [12842, 12880], [12896, 12927], [12938, 12976], [12992, 13311], 256 | [19894, 19967], [40908, 40959], [42125, 42191], [42238, 42239], [42509, 42511], 257 | [42540, 42559], [42592, 42593], [42607, 42622], [42648, 42655], [42736, 42774], 258 | [42784, 42785], [42889, 42890], [42893, 43002], [43043, 43055], [43062, 43071], 259 | [43124, 43137], [43188, 43215], [43226, 43249], [43256, 43258], [43260, 43263], 260 | [43302, 43311], [43335, 43359], [43389, 43395], [43443, 43470], [43482, 43519], 261 | [43561, 43583], [43596, 43599], [43610, 43615], [43639, 43641], [43643, 43647], 262 | [43698, 43700], [43703, 43704], [43710, 43711], [43715, 43738], [43742, 43967], 263 | [44003, 44015], [44026, 44031], [55204, 55215], [55239, 55242], [55292, 55295], 264 | [57344, 63743], [64046, 64047], [64110, 64111], [64218, 64255], [64263, 64274], 265 | [64280, 64284], [64434, 64466], [64830, 64847], [64912, 64913], [64968, 65007], 266 | [65020, 65135], [65277, 65295], [65306, 65312], [65339, 65344], [65371, 65381], 267 | [65471, 65473], [65480, 65481], [65488, 65489], [65496, 65497]]; 268 | for (i = 0; i < ranges.length; i++) { 269 | start = ranges[i][0]; 270 | end = ranges[i][1]; 271 | for (j = start; j <= end; j++) { 272 | result[j] = true; 273 | } 274 | } 275 | return result; 276 | })(); 277 | 278 | function splitQuery(query) { 279 | var result = []; 280 | var start = -1; 281 | for (var i = 0; i < query.length; i++) { 282 | if (splitChars[query.charCodeAt(i)]) { 283 | if (start !== -1) { 284 | result.push(query.slice(start, i)); 285 | start = -1; 286 | } 287 | } else if (start === -1) { 288 | start = i; 289 | } 290 | } 291 | if (start !== -1) { 292 | result.push(query.slice(start)); 293 | } 294 | return result; 295 | } 296 | 297 | 298 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/minus.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/html/_static/minus.png -------------------------------------------------------------------------------- /ds_book/_build/html/_static/mystnb.css: -------------------------------------------------------------------------------- 1 | /* Whole cell */ 2 | div.container.cell { 3 | padding-left: 0; 4 | margin-bottom: 1em; 5 | } 6 | 7 | /* Removing all background formatting so we can control at the div level */ 8 | .cell_input div.highlight, .cell_input pre, .cell_output .output * { 9 | border: none; 10 | box-shadow: none; 11 | } 12 | 13 | .cell_output .output pre, .cell_input pre { 14 | margin: 0px; 15 | } 16 | 17 | /* Input cells */ 18 | div.cell div.cell_input { 19 | padding-left: 0em; 20 | padding-right: 0em; 21 | border: 1px #ccc solid; 22 | background-color: #f7f7f7; 23 | border-left-color: green; 24 | border-left-width: medium; 25 | } 26 | 27 | div.cell_input > div, div.cell_output div.output > div.highlight { 28 | margin: 0em !important; 29 | border: none !important; 30 | } 31 | 32 | /* All cell outputs */ 33 | .cell_output { 34 | padding-left: 1em; 35 | padding-right: 0em; 36 | margin-top: 1em; 37 | } 38 | 39 | /* Outputs from jupyter_sphinx overrides to remove extra CSS */ 40 | div.section div.jupyter_container { 41 | padding: .4em; 42 | margin: 0 0 .4em 0; 43 | background-color: none; 44 | border: none; 45 | -moz-box-shadow: none; 46 | -webkit-box-shadow: none; 47 | box-shadow: none; 48 | } 49 | 50 | /* Text outputs from cells */ 51 | .cell_output .output.text_plain, 52 | .cell_output .output.traceback, 53 | .cell_output .output.stream, 54 | .cell_output .output.stderr 55 | { 56 | background: #fcfcfc; 57 | margin-top: 1em; 58 | margin-bottom: 0em; 59 | box-shadow: none; 60 | } 61 | 62 | .cell_output .output.text_plain, 63 | .cell_output .output.stream, 64 | .cell_output .output.stderr { 65 | border: 1px solid #f7f7f7; 66 | } 67 | 68 | .cell_output .output.stderr { 69 | background: #fdd; 70 | } 71 | 72 | .cell_output .output.traceback { 73 | border: 1px solid #ffd6d6; 74 | } 75 | 76 | /* Math align to the left */ 77 | .cell_output .MathJax_Display { 78 | text-align: left !important; 79 | } 80 | 81 | /* Pandas tables. Pulled from the Jupyter / nbsphinx CSS */ 82 | div.cell_output table { 83 | border: none; 84 | border-collapse: collapse; 85 | border-spacing: 0; 86 | color: black; 87 | font-size: 1em; 88 | table-layout: fixed; 89 | } 90 | div.cell_output thead { 91 | border-bottom: 1px solid black; 92 | vertical-align: bottom; 93 | } 94 | div.cell_output tr, 95 | div.cell_output th, 96 | div.cell_output td { 97 | text-align: right; 98 | vertical-align: middle; 99 | padding: 0.5em 0.5em; 100 | line-height: normal; 101 | white-space: normal; 102 | max-width: none; 103 | border: none; 104 | } 105 | div.cell_output th { 106 | font-weight: bold; 107 | } 108 | div.cell_output tbody tr:nth-child(odd) { 109 | background: #f5f5f5; 110 | } 111 | div.cell_output tbody tr:hover { 112 | background: rgba(66, 165, 245, 0.2); 113 | } 114 | 115 | 116 | /* Inline text from `paste` operation */ 117 | 118 | span.pasted-text { 119 | font-weight: bold; 120 | } 121 | 122 | span.pasted-inline img { 123 | max-height: 2em; 124 | } 125 | 126 | tbody span.pasted-inline img { 127 | max-height: none; 128 | } 129 | 130 | /* Font colors for translated ANSI escape sequences 131 | Color values are adapted from share/jupyter/nbconvert/templates/classic/static/style.css 132 | */ 133 | div.highlight .-Color-Bold { 134 | font-weight: bold; 135 | } 136 | div.highlight .-Color[class*=-Black] { 137 | color :#3E424D 138 | } 139 | div.highlight .-Color[class*=-Red] { 140 | color: #E75C58 141 | } 142 | div.highlight .-Color[class*=-Green] { 143 | color: #00A250 144 | } 145 | div.highlight .-Color[class*=-Yellow] { 146 | color: yellow 147 | } 148 | div.highlight .-Color[class*=-Blue] { 149 | color: #208FFB 150 | } 151 | div.highlight .-Color[class*=-Magenta] { 152 | color: #D160C4 153 | } 154 | div.highlight .-Color[class*=-Cyan] { 155 | color: #60C6C8 156 | } 157 | div.highlight .-Color[class*=-White] { 158 | color: #C5C1B4 159 | } 160 | div.highlight .-Color[class*=-BGBlack] { 161 | background-color: #3E424D 162 | } 163 | div.highlight .-Color[class*=-BGRed] { 164 | background-color: #E75C58 165 | } 166 | div.highlight .-Color[class*=-BGGreen] { 167 | background-color: #00A250 168 | } 169 | div.highlight .-Color[class*=-BGYellow] { 170 | background-color: yellow 171 | } 172 | div.highlight .-Color[class*=-BGBlue] { 173 | background-color: #208FFB 174 | } 175 | div.highlight .-Color[class*=-BGMagenta] { 176 | background-color: #D160C4 177 | } 178 | div.highlight .-Color[class*=-BGCyan] { 179 | background-color: #60C6C8 180 | } 181 | div.highlight .-Color[class*=-BGWhite] { 182 | background-color: #C5C1B4 183 | } 184 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/panels-main.c949a650a448cc0ae9fd3441c0e17fb0.css: -------------------------------------------------------------------------------- 1 | details.dropdown .summary-title{padding-right:3em !important;-moz-user-select:none;-ms-user-select:none;-webkit-user-select:none;user-select:none}details.dropdown:hover{cursor:pointer}details.dropdown .summary-content{cursor:default}details.dropdown summary{list-style:none;padding:1em}details.dropdown summary .octicon.no-title{vertical-align:middle}details.dropdown[open] summary .octicon.no-title{visibility:hidden}details.dropdown summary::-webkit-details-marker{display:none}details.dropdown summary:focus{outline:none}details.dropdown summary:hover .summary-up svg,details.dropdown summary:hover .summary-down svg{opacity:1}details.dropdown .summary-up svg,details.dropdown .summary-down svg{display:block;opacity:.6}details.dropdown .summary-up,details.dropdown .summary-down{pointer-events:none;position:absolute;right:1em;top:.75em}details.dropdown[open] .summary-down{visibility:hidden}details.dropdown:not([open]) .summary-up{visibility:hidden}details.dropdown.fade-in[open] summary~*{-moz-animation:panels-fade-in .5s ease-in-out;-webkit-animation:panels-fade-in .5s ease-in-out;animation:panels-fade-in .5s ease-in-out}details.dropdown.fade-in-slide-down[open] summary~*{-moz-animation:panels-fade-in .5s ease-in-out, panels-slide-down .5s ease-in-out;-webkit-animation:panels-fade-in .5s ease-in-out, panels-slide-down .5s ease-in-out;animation:panels-fade-in .5s ease-in-out, panels-slide-down .5s ease-in-out}@keyframes panels-fade-in{0%{opacity:0}100%{opacity:1}}@keyframes panels-slide-down{0%{transform:translate(0, -10px)}100%{transform:translate(0, 0)}}.octicon{display:inline-block;fill:currentColor;vertical-align:text-top}.tabbed-content{box-shadow:0 -.0625rem var(--tabs-color-overline),0 .0625rem var(--tabs-color-underline);display:none;order:99;padding-bottom:.75rem;padding-top:.75rem;width:100%}.tabbed-content>:first-child{margin-top:0 !important}.tabbed-content>:last-child{margin-bottom:0 !important}.tabbed-content>.tabbed-set{margin:0}.tabbed-set{border-radius:.125rem;display:flex;flex-wrap:wrap;margin:1em 0;position:relative}.tabbed-set>input{opacity:0;position:absolute}.tabbed-set>input:checked+label{border-color:var(--tabs-color-label-active);color:var(--tabs-color-label-active)}.tabbed-set>input:checked+label+.tabbed-content{display:block}.tabbed-set>input:focus+label{outline-style:auto}.tabbed-set>input:not(.focus-visible)+label{outline:none;-webkit-tap-highlight-color:transparent}.tabbed-set>label{border-bottom:.125rem solid transparent;color:var(--tabs-color-label-inactive);cursor:pointer;font-size:var(--tabs-size-label);font-weight:700;padding:1em 1.25em .5em;transition:color 250ms;width:auto;z-index:1}html .tabbed-set>label:hover{color:var(--tabs-color-label-active)} 2 | -------------------------------------------------------------------------------- /ds_book/_build/html/_static/panels-variables.06eb56fa6e07937060861dad626602ad.css: -------------------------------------------------------------------------------- 1 | :root { 2 | --tabs-color-label-active: hsla(231, 99%, 66%, 1); 3 | --tabs-color-label-inactive: rgba(178, 206, 245, 0.62); 4 | --tabs-color-overline: rgb(207, 236, 238); 5 | --tabs-color-underline: rgb(207, 236, 238); 6 | --tabs-size-label: 1rem; 7 | } -------------------------------------------------------------------------------- /ds_book/_build/html/_static/plus.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/html/_static/plus.png -------------------------------------------------------------------------------- /ds_book/_build/html/_static/pygments.css: -------------------------------------------------------------------------------- 1 | pre { line-height: 125%; } 2 | td.linenos .normal { color: inherit; background-color: transparent; padding-left: 5px; padding-right: 5px; } 3 | span.linenos { color: inherit; background-color: transparent; padding-left: 5px; padding-right: 5px; } 4 | td.linenos .special { color: #000000; background-color: #ffffc0; padding-left: 5px; padding-right: 5px; } 5 | span.linenos.special { color: #000000; background-color: #ffffc0; padding-left: 5px; padding-right: 5px; } 6 | .highlight .hll { background-color: #ffffcc } 7 | .highlight { background: #eeffcc; } 8 | .highlight .c { color: #408090; font-style: italic } /* Comment */ 9 | .highlight .err { border: 1px solid #FF0000 } /* Error */ 10 | .highlight .k { color: #007020; font-weight: bold } /* Keyword */ 11 | .highlight .o { color: #666666 } /* Operator */ 12 | .highlight .ch { color: #408090; font-style: italic } /* Comment.Hashbang */ 13 | .highlight .cm { color: #408090; font-style: italic } /* Comment.Multiline */ 14 | .highlight .cp { color: #007020 } /* Comment.Preproc */ 15 | .highlight .cpf { color: #408090; font-style: italic } /* Comment.PreprocFile */ 16 | .highlight .c1 { color: #408090; font-style: italic } /* Comment.Single */ 17 | .highlight .cs { color: #408090; background-color: #fff0f0 } /* Comment.Special */ 18 | .highlight .gd { color: #A00000 } /* Generic.Deleted */ 19 | .highlight .ge { font-style: italic } /* Generic.Emph */ 20 | .highlight .gr { color: #FF0000 } /* Generic.Error */ 21 | .highlight .gh { color: #000080; font-weight: bold } /* Generic.Heading */ 22 | .highlight .gi { color: #00A000 } /* Generic.Inserted */ 23 | .highlight .go { color: #333333 } /* Generic.Output */ 24 | .highlight .gp { color: #c65d09; font-weight: bold } /* Generic.Prompt */ 25 | .highlight .gs { font-weight: bold } /* Generic.Strong */ 26 | .highlight .gu { color: #800080; font-weight: bold } /* Generic.Subheading */ 27 | .highlight .gt { color: #0044DD } /* Generic.Traceback */ 28 | .highlight .kc { color: #007020; font-weight: bold } /* Keyword.Constant */ 29 | .highlight .kd { color: #007020; font-weight: bold } /* Keyword.Declaration */ 30 | .highlight .kn { color: #007020; font-weight: bold } /* Keyword.Namespace */ 31 | .highlight .kp { color: #007020 } /* Keyword.Pseudo */ 32 | .highlight .kr { color: #007020; font-weight: bold } /* Keyword.Reserved */ 33 | .highlight .kt { color: #902000 } /* Keyword.Type */ 34 | .highlight .m { color: #208050 } /* Literal.Number */ 35 | .highlight .s { color: #4070a0 } /* Literal.String */ 36 | .highlight .na { color: #4070a0 } /* Name.Attribute */ 37 | .highlight .nb { color: #007020 } /* Name.Builtin */ 38 | .highlight .nc { color: #0e84b5; font-weight: bold } /* Name.Class */ 39 | .highlight .no { color: #60add5 } /* Name.Constant */ 40 | .highlight .nd { color: #555555; font-weight: bold } /* Name.Decorator */ 41 | .highlight .ni { color: #d55537; font-weight: bold } /* Name.Entity */ 42 | .highlight .ne { color: #007020 } /* Name.Exception */ 43 | .highlight .nf { color: #06287e } /* Name.Function */ 44 | .highlight .nl { color: #002070; font-weight: bold } /* Name.Label */ 45 | .highlight .nn { color: #0e84b5; font-weight: bold } /* Name.Namespace */ 46 | .highlight .nt { color: #062873; font-weight: bold } /* Name.Tag */ 47 | .highlight .nv { color: #bb60d5 } /* Name.Variable */ 48 | .highlight .ow { color: #007020; font-weight: bold } /* Operator.Word */ 49 | .highlight .w { color: #bbbbbb } /* Text.Whitespace */ 50 | .highlight .mb { color: #208050 } /* Literal.Number.Bin */ 51 | .highlight .mf { color: #208050 } /* Literal.Number.Float */ 52 | .highlight .mh { color: #208050 } /* Literal.Number.Hex */ 53 | .highlight .mi { color: #208050 } /* Literal.Number.Integer */ 54 | .highlight .mo { color: #208050 } /* Literal.Number.Oct */ 55 | .highlight .sa { color: #4070a0 } /* Literal.String.Affix */ 56 | .highlight .sb { color: #4070a0 } /* Literal.String.Backtick */ 57 | .highlight .sc { color: #4070a0 } /* Literal.String.Char */ 58 | .highlight .dl { color: #4070a0 } /* Literal.String.Delimiter */ 59 | .highlight .sd { color: #4070a0; font-style: italic } /* Literal.String.Doc */ 60 | .highlight .s2 { color: #4070a0 } /* Literal.String.Double */ 61 | .highlight .se { color: #4070a0; font-weight: bold } /* Literal.String.Escape */ 62 | .highlight .sh { color: #4070a0 } /* Literal.String.Heredoc */ 63 | .highlight .si { color: #70a0d0; font-style: italic } /* Literal.String.Interpol */ 64 | .highlight .sx { color: #c65d09 } /* Literal.String.Other */ 65 | .highlight .sr { color: #235388 } /* Literal.String.Regex */ 66 | .highlight .s1 { color: #4070a0 } /* Literal.String.Single */ 67 | .highlight .ss { color: #517918 } /* Literal.String.Symbol */ 68 | .highlight .bp { color: #007020 } /* Name.Builtin.Pseudo */ 69 | .highlight .fm { color: #06287e } /* Name.Function.Magic */ 70 | .highlight .vc { color: #bb60d5 } /* Name.Variable.Class */ 71 | .highlight .vg { color: #bb60d5 } /* Name.Variable.Global */ 72 | .highlight .vi { color: #bb60d5 } /* Name.Variable.Instance */ 73 | .highlight .vm { color: #bb60d5 } /* Name.Variable.Magic */ 74 | .highlight .il { color: #208050 } /* Literal.Number.Integer.Long */ -------------------------------------------------------------------------------- /ds_book/_build/html/_static/sphinx-book-theme.12a9622fbb08dcb3a2a40b2c02b83a57.js: -------------------------------------------------------------------------------- 1 | var initTriggerNavBar=()=>{if($(window).width()<768){$("#navbar-toggler").trigger("click")}} 2 | var scrollToActive=()=>{var navbar=document.getElementById('site-navigation') 3 | var active_pages=navbar.querySelectorAll(".active") 4 | var active_page=active_pages[active_pages.length-1] 5 | if(active_page!==undefined&&active_page.offsetTop>($(window).height()*.5)){navbar.scrollTop=active_page.offsetTop-($(window).height()*.2)}} 6 | var sbRunWhenDOMLoaded=cb=>{if(document.readyState!='loading'){cb()}else if(document.addEventListener){document.addEventListener('DOMContentLoaded',cb)}else{document.attachEvent('onreadystatechange',function(){if(document.readyState=='complete')cb()})}} 7 | function toggleFullScreen(){var navToggler=$("#navbar-toggler");if(!document.fullscreenElement){document.documentElement.requestFullscreen();if(!navToggler.hasClass("collapsed")){navToggler.click();}}else{if(document.exitFullscreen){document.exitFullscreen();if(navToggler.hasClass("collapsed")){navToggler.click();}}}} 8 | var initTooltips=()=>{$(document).ready(function(){$('[data-toggle="tooltip"]').tooltip();});} 9 | var initTocHide=()=>{var scrollTimeout;var throttle=200;var tocHeight=$("#bd-toc-nav").outerHeight(true)+$(".bd-toc").outerHeight(true);var hideTocAfter=tocHeight+200;var checkTocScroll=function(){var margin_content=$(".margin, .tag_margin, .full-width, .full_width, .tag_full-width, .tag_full_width, .sidebar, .tag_sidebar, .popout, .tag_popout");margin_content.each((index,item)=>{var topOffset=$(item).offset().top-$(window).scrollTop();var bottomOffset=topOffset+$(item).outerHeight(true);var topOverlaps=((topOffset>=0)&&(topOffset205 | Please activate JavaScript to enable the search 206 | functionality. 207 |
208 |210 | Searching for multiple words only shows matches that contain 211 | all words. 212 |
213 | 218 | 219 |\n",
68 | "\n",
69 | "TensorFlow graph example (from [https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4](https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4)).\n",
70 | ":::\n",
71 | "\n",
72 | "\n",
73 | "#### Why do so many people like TensorFlow?\n",
74 | "TensorFlow is intentionally user-friendly, with helpful plugins to visualize model training and a useful software debugging tool. As well, TensorFlow is highly scalable, with easy deployment on both CPUs and GPUs."
75 | ]
76 | },
77 | {
78 | "cell_type": "markdown",
79 | "metadata": {},
80 | "source": [
81 | "### What is Keras?\n",
82 | "\n",
83 | "[Keras](https://keras.io/about/) is an API built on Python which reduces the cognitive load associated with programming models through human readability and simple and consistent structures.\n",
84 | "\n",
85 | "Keras is what some might call a wrapper for TensorFlow. It is intended for rapid experimentation.\n",
86 | "\n",
87 | "Tha main components of Keras include:\n",
88 | "1. A models API, which enables one to construct a model with varying levels of complexity depending on use case. We will use the [Functional API](https://keras.io/guides/functional_api/) subclass.\n",
89 | "2. A layers API, which allows one to define the tensor in/tensor out computation functions.\n",
90 | "3. A callback API, which enables one to program specific actions to occur during training, such as log training metrics, visualize interim/internal states and statistics of the model during training, and perform early stopping when the model converges.\n",
91 | "4. A data preprocessing API, which offers support for prepping raw data from disk to model ready Tensor format.\n",
92 | "5. An optimizer API where all of the state of the art optimizers can be plugged in. Learning rate decay / scheduling can also be implemented as part of this API.\n",
93 | "6. A metrics API which is used for assessing the performance of the model during training. A metric is the target to optimize during training, with specific metrics chosen for specific modeling objectives.\n",
94 | "7. A loss API that informs the model quantitatively how much it should try to minimize during training by providing a measure of error. Similar to metrics, specific loss functions are selected for specific modeling objectives.\n",
95 | "\n",
96 | "With the Functional API, our main workflow will follow the diagram below.\n",
97 | "\n",
98 | ":::{figure-md} Keras-fig\n",
99 | "\n",
100 | "\n",
101 | "Keras Functional API diagram (from [https://miro.com/app/board/o9J_lhnKhVE=/](hhttps://miro.com/app/board/o9J_lhnKhVE=/)).\n",
102 | ":::"
103 | ]
104 | }
105 | ],
106 | "metadata": {
107 | "kernelspec": {
108 | "display_name": "Python 3",
109 | "language": "python",
110 | "name": "python3"
111 | },
112 | "language_info": {
113 | "codemirror_mode": {
114 | "name": "ipython",
115 | "version": 3
116 | },
117 | "file_extension": ".py",
118 | "mimetype": "text/x-python",
119 | "name": "python",
120 | "nbconvert_exporter": "python",
121 | "pygments_lexer": "ipython3",
122 | "version": "3.8.11"
123 | }
124 | },
125 | "nbformat": 4,
126 | "nbformat_minor": 2
127 | }
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson1b_Intro_TensorFlow_Keras.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python
2 | # coding: utf-8
3 |
4 | # # Introduction to TensorFlow and Keras
5 |
6 | # ## Objectives
7 |
8 | # The goal of this notebook is to teach some basics of the TensorFlow framework and the Keras API.
9 |
10 | # ### What is TensorFlow?
11 | #
12 | # [TensorFlow](https://www.tensorflow.org/guide]) is an open-source framework developed in late 2015 by Google for building various machine learning and deep learning models. TensorFlow is free and open-source, thanks to the Apache Open Source license.
13 | #
14 | # The main objective of using TensorFlow is to reduce the complexity of implementing computations on large numerical data sets. In practice, these large computations can manifest as training and inference with machine learning or deep learning models.
15 | #
16 | # TensorFlow was designed to operate with multiple CPUs or GPUs, as well as a growing number of mobile operating systems. The framework includes wrappers in Python, C++, and Java.
17 | #
18 | # #### How does it work?
19 | # TensorFlow accepts inputs as a multi-dimensional array called a Tensor, which allows the programmer to create dataflow graphs and structures specifying how data travels through. The framework is designed to support creation of a flowchart of operations to be applied to input Tensors, which travel in one direction and out the other.
20 | #
21 | # #### TensorFlow's structure
22 | # There are three main components to TensorFlow's structure.
23 | #
24 | # 1. preprocessing the data
25 | # 2. building the model
26 | # 3. training and estimating the model
27 | #
28 | # The name Tensorflow derives from the way in which the framework receives input in the form of a multi-dimensional array, i.e. the tensors. These tensors travel sequentially through the specified flowchart of the operations, entering at one end and culminating as output at the other end.
29 | #
30 | # #### What are the key TensorFlow components?
31 | # **Tensor**
32 | #
33 | # A tensor is an n-dimensional vector or matrix. In theory, a tensor may represent any form of data. The values belonging to a tensor all share the same data type and often the same shape / dimensionality. A tensor can describe the input data and the output of a calculation. The main difference between a tensor and a conventional n-dimensional array is that tensors are immutable.
34 | #
35 | #
36 | # **Graphs**
37 | #
38 | # TensorFlow uses a graph framework. Graphs collect and summarize all of the calculations and offer several benefits:
39 | #
40 | # 1. They are designed to work on CPUs or GPUs, as well as on mobile devices.
41 | # 2. Graphs are portable which enables the computations to be saved for immediate or later usage. Otherwise stated, the graph can be frozen and run at a later time.
42 | # 3. Graph calculations are executed by linking tensors together.
43 | # 4. For each tensor, there is a node and an edge. The node carries out the mathematical process and produces endpoint outputs. The input/output connections are represented by the edges.
44 | # 5. All nodes are linked together, so the graph itself is a depiction of the operations and relationships that exist between the nodes.
45 | #
46 | # In TensorFlow, all operations are carried out within a graph, which in effect is a series of computations that happen in order. Each individual operation is referred to as an op node.
47 | #
48 | # :::{figure-md} TFgraph-fig
49 | # 
50 | #
51 | # TensorFlow graph example (from [https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4](https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4)).
52 | # :::
53 | #
54 | #
55 | # #### Why do so many people like TensorFlow?
56 | # TensorFlow is intentionally user-friendly, with helpful plugins to visualize model training and a useful software debugging tool. As well, TensorFlow is highly scalable, with easy deployment on both CPUs and GPUs.
57 |
58 | # ### What is Keras?
59 | #
60 | # [Keras](https://keras.io/about/) is an API built on Python which reduces the cognitive load associated with programming models through human readability and simple and consistent structures.
61 | #
62 | # Keras is what some might call a wrapper for TensorFlow. It is intended for rapid experimentation.
63 | #
64 | # Tha main components of Keras include:
65 | # 1. A models API, which enables one to construct a model with varying levels of complexity depending on use case. We will use the [Functional API](https://keras.io/guides/functional_api/) subclass.
66 | # 2. A layers API, which allows one to define the tensor in/tensor out computation functions.
67 | # 3. A callback API, which enables one to program specific actions to occur during training, such as log training metrics, visualize interim/internal states and statistics of the model during training, and perform early stopping when the model converges.
68 | # 4. A data preprocessing API, which offers support for prepping raw data from disk to model ready Tensor format.
69 | # 5. An optimizer API where all of the state of the art optimizers can be plugged in. Learning rate decay / scheduling can also be implemented as part of this API.
70 | # 6. A metrics API which is used for assessing the performance of the model during training. A metric is the target to optimize during training, with specific metrics chosen for specific modeling objectives.
71 | # 7. A loss API that informs the model quantitatively how much it should try to minimize during training by providing a measure of error. Similar to metrics, specific loss functions are selected for specific modeling objectives.
72 | #
73 | # With the Functional API, our main workflow will follow the diagram below.
74 | #
75 | # :::{figure-md} Keras-fig
76 | # 
77 | #
78 | # Keras Functional API diagram (from [https://miro.com/app/board/o9J_lhnKhVE=/](hhttps://miro.com/app/board/o9J_lhnKhVE=/)).
79 | # :::
80 |
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson1b_annex.ipynb:
--------------------------------------------------------------------------------
1 | {
2 | "cells": [
3 | {
4 | "cell_type": "markdown",
5 | "metadata": {},
6 | "source": [
7 | "# Introduction to TensorFlow and Keras"
8 | ]
9 | },
10 | {
11 | "cell_type": "markdown",
12 | "metadata": {},
13 | "source": [
14 | "## Objectives"
15 | ]
16 | },
17 | {
18 | "cell_type": "markdown",
19 | "metadata": {},
20 | "source": [
21 | "The goal of this notebook is to teach some basics of the TensorFlow framework and the Keras API."
22 | ]
23 | },
24 | {
25 | "cell_type": "markdown",
26 | "metadata": {},
27 | "source": [
28 | "### What is TensorFlow?\n",
29 | "\n",
30 | "[TensorFlow](https://www.tensorflow.org/guide]) is an open-source framework developed by Google for building various machine learning and deep learning models. Although originally released in late 2015, the first stable version arrived in 2017. TensorFlow is free and open-source, thanks to the Apache Open Source license.\n",
31 | "\n",
32 | "The main objective of using TensorFlow is to reduce the complexity of implementing computations on large numerical data sets. In practice, these large computations can manifest as training and inference with machine learning or deep learning models.\n",
33 | "\n",
34 | "TensorFlow was designed to operate with multiple CPUs or GPUs, as well as a growing number of mobile operating systems. The framework includes wrappers in Python, C++, and Java.\n",
35 | "\n",
36 | "#### How does it work?\n",
37 | "TensorFlow accepts inputs as a multi-dimensional array called a Tensor, which allows the programmer to create dataflow graphs and structures specifying how data travels through. The framework is designed to support creation of a flowchart of operations to be applied to input Tensors, which travel in one direction and out the other.\n",
38 | "\n",
39 | "#### TensorFlow’s structure\n",
40 | "There are three main components to TensorFlow's structure.\n",
41 | "\n",
42 | "1. preprocessing the data\n",
43 | "2. building the model\n",
44 | "3. training and estimating the model\n",
45 | "\n",
46 | "The name TensorFlow derives from the way in which the framework receives input in the form of a multi-dimensional array, i.e. the tensors. These tensors travel sequentially through the specified flowchart of the operations, entering at one end and culminate as output at the other end. \n",
47 | "\n",
48 | "#### What are TensorFlow components?\n",
49 | "**Tensor**\n",
50 | "\n",
51 | "Tensors, the basic unit of data in this framework, are involved in every computation of TensorFlow. A tensor is an n-dimensional vector or matrix. In theory, a tensor may represent any form of data. The values belonging to a tensor all share the same data type and often the same shape / dimensionality. A tensor can describe the input data and the output of a calculation. \n",
52 | "\n",
53 | "In TensorFlow, all operations are carried out within a graph, which in effect is a series of computations that happen in order. Each indivisual operation is referred to as an op node. \n",
54 | "\n",
55 | "**Graphs**\n",
56 | "\n",
57 | "TensorFlow uses a graph framework. Graphs collect and summarize all of the calculations and offer several benefits:\n",
58 | "\n",
59 | "1. They are designed to work on CPUs or GPUs, as well as on mobile devices.\n",
60 | "2. Graphs are portable which enables the computations to be saved for immediate or later usage. Otherwsie stated, the graph can be frozen and run at a later time.\n",
61 | "3. Graph calculations are executed by linking tensors together.\n",
62 | "4. For each tensor, there is a node and an edge. The node carries out the mathematical process and produces endpoint outputs. The input/output connections are represented by the edges.\n",
63 | "5. All nodes are linked together, so the graph itself is a depiction of the operations and relationships that exist between the nodes.\n",
64 | "\n",
65 | ":::{figure-md} TFgraph-fig\n",
66 | "\n",
67 | "\n",
68 | "TensorFlow graph example (from [https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4](https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4)).\n",
69 | ":::\n",
70 | "\n",
71 | "#### Why do so many people like TensorFlow?\n",
72 | "TensorFlow is intentionally user-friendly, with helpful plugins to visualize model training and a useful software debugging tool. As well, TensorFlow is highly scalable, with easy deployment on both CPUs and GPUs."
73 | ]
74 | },
75 | {
76 | "cell_type": "markdown",
77 | "metadata": {},
78 | "source": [
79 | "### What is Keras?\n",
80 | "\n",
81 | "[Keras](https://keras.io/about/) is an API built on Python, with human readability at the forefront of its design. With simple and consistent structures and methods extensible across many machine learning and deep learning applications, Keras reduces the cognitive load associated with programming models. Furthermore, the API seeks to minimize the need for programmer interaction by abstracting many complexities into easily callable functions. Lastly, Keras features clear & actionable error messaging, complemented by comprehensive and digestible documentation and developer guides.\n",
82 | "\n",
83 | "Keras is what some might call a wrapper for TensorFlow. By that, one means to say that Keras simplifies a programmer's interaction with TensorFlow through refinement of key methods and constructs.\n",
84 | "\n",
85 | "Importantly, Keras is intended for rapid experimentation.\n",
86 | "\n",
87 | "Tha main components of Keras include:\n",
88 | "1. A models API, which enables one to construct a model with varying levels of complexity depending on use case. We will use the [Functional API](https://keras.io/guides/functional_api/) subclass.\n",
89 | "2. A layers API, which allows one to define the tensor in/tnesor out computation functions.\n",
90 | "3. A callback API, which enables one to program specific actions to occur during training, such as log training metrics, visualize interim/internal states and statistics of the model during training, and perform early stopping when the model converges.\n",
91 | "4. A data preprocessing API, which offers support for prepping raw data from disk to model ready Tensor format.\n",
92 | "5. An optimizer API where all of the state of the art optimizers can be plugged in. Learning rate decay / scheduling can also be implemented a spart of this API.\n",
93 | "6. A metrics API which is used for assessing the performance of the model during training. A metric is the target to optimize during training, with specific metrics chosen for specific modeling objectives.\n",
94 | "7. A loss API that informs the model quantitatively how much it should try to minimize during training by providing a measure of error. Similar to metrics, specific loss functions are selected for specific modelung objectives.\n",
95 | "\n",
96 | "With the Functional API, our main workflow will follow the diagram below.\n",
97 | "\n",
98 | ":::{figure-md} Keras-fig\n",
99 | "\n",
100 | "\n",
101 | "Keras Functional API diagram (from [https://miro.com/app/board/o9J_lhnKhVE=/](hhttps://miro.com/app/board/o9J_lhnKhVE=/)).\n",
102 | ":::"
103 | ]
104 | }
105 | ],
106 | "metadata": {
107 | "kernelspec": {
108 | "display_name": "Python 3",
109 | "language": "python",
110 | "name": "python3"
111 | },
112 | "language_info": {
113 | "codemirror_mode": {
114 | "name": "ipython",
115 | "version": 3
116 | },
117 | "file_extension": ".py",
118 | "mimetype": "text/x-python",
119 | "name": "python",
120 | "nbconvert_exporter": "python",
121 | "pygments_lexer": "ipython3",
122 | "version": "3.8.11"
123 | }
124 | },
125 | "nbformat": 4,
126 | "nbformat_minor": 2
127 | }
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson1b_annex.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python
2 | # coding: utf-8
3 |
4 | # # Introduction to TensorFlow and Keras
5 |
6 | # ## Objectives
7 |
8 | # The goal of this notebook is to teach some basics of the TensorFlow framework and the Keras API.
9 |
10 | # ### What is TensorFlow?
11 | #
12 | # [TensorFlow](https://www.tensorflow.org/guide]) is an open-source framework developed by Google for building various machine learning and deep learning models. Although originally released in late 2015, the first stable version arrived in 2017. TensorFlow is free and open-source, thanks to the Apache Open Source license.
13 | #
14 | # The main objective of using TensorFlow is to reduce the complexity of implementing computations on large numerical data sets. In practice, these large computations can manifest as training and inference with machine learning or deep learning models.
15 | #
16 | # TensorFlow was designed to operate with multiple CPUs or GPUs, as well as a growing number of mobile operating systems. The framework includes wrappers in Python, C++, and Java.
17 | #
18 | # #### How does it work?
19 | # TensorFlow accepts inputs as a multi-dimensional array called a Tensor, which allows the programmer to create dataflow graphs and structures specifying how data travels through. The framework is designed to support creation of a flowchart of operations to be applied to input Tensors, which travel in one direction and out the other.
20 | #
21 | # #### TensorFlow’s structure
22 | # There are three main components to TensorFlow's structure.
23 | #
24 | # 1. preprocessing the data
25 | # 2. building the model
26 | # 3. training and estimating the model
27 | #
28 | # The name TensorFlow derives from the way in which the framework receives input in the form of a multi-dimensional array, i.e. the tensors. These tensors travel sequentially through the specified flowchart of the operations, entering at one end and culminate as output at the other end.
29 | #
30 | # #### What are TensorFlow components?
31 | # **Tensor**
32 | #
33 | # Tensors, the basic unit of data in this framework, are involved in every computation of TensorFlow. A tensor is an n-dimensional vector or matrix. In theory, a tensor may represent any form of data. The values belonging to a tensor all share the same data type and often the same shape / dimensionality. A tensor can describe the input data and the output of a calculation.
34 | #
35 | # In TensorFlow, all operations are carried out within a graph, which in effect is a series of computations that happen in order. Each indivisual operation is referred to as an op node.
36 | #
37 | # **Graphs**
38 | #
39 | # TensorFlow uses a graph framework. Graphs collect and summarize all of the calculations and offer several benefits:
40 | #
41 | # 1. They are designed to work on CPUs or GPUs, as well as on mobile devices.
42 | # 2. Graphs are portable which enables the computations to be saved for immediate or later usage. Otherwsie stated, the graph can be frozen and run at a later time.
43 | # 3. Graph calculations are executed by linking tensors together.
44 | # 4. For each tensor, there is a node and an edge. The node carries out the mathematical process and produces endpoint outputs. The input/output connections are represented by the edges.
45 | # 5. All nodes are linked together, so the graph itself is a depiction of the operations and relationships that exist between the nodes.
46 | #
47 | # :::{figure-md} TFgraph-fig
48 | #
49 | #
50 | # TensorFlow graph example (from [https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4](https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4)).
51 | # :::
52 | #
53 | # #### Why do so many people like TensorFlow?
54 | # TensorFlow is intentionally user-friendly, with helpful plugins to visualize model training and a useful software debugging tool. As well, TensorFlow is highly scalable, with easy deployment on both CPUs and GPUs.
55 |
56 | # ### What is Keras?
57 | #
58 | # [Keras](https://keras.io/about/) is an API built on Python, with human readability at the forefront of its design. With simple and consistent structures and methods extensible across many machine learning and deep learning applications, Keras reduces the cognitive load associated with programming models. Furthermore, the API seeks to minimize the need for programmer interaction by abstracting many complexities into easily callable functions. Lastly, Keras features clear & actionable error messaging, complemented by comprehensive and digestible documentation and developer guides.
59 | #
60 | # Keras is what some might call a wrapper for TensorFlow. By that, one means to say that Keras simplifies a programmer's interaction with TensorFlow through refinement of key methods and constructs.
61 | #
62 | # Importantly, Keras is intended for rapid experimentation.
63 | #
64 | # Tha main components of Keras include:
65 | # 1. A models API, which enables one to construct a model with varying levels of complexity depending on use case. We will use the [Functional API](https://keras.io/guides/functional_api/) subclass.
66 | # 2. A layers API, which allows one to define the tensor in/tnesor out computation functions.
67 | # 3. A callback API, which enables one to program specific actions to occur during training, such as log training metrics, visualize interim/internal states and statistics of the model during training, and perform early stopping when the model converges.
68 | # 4. A data preprocessing API, which offers support for prepping raw data from disk to model ready Tensor format.
69 | # 5. An optimizer API where all of the state of the art optimizers can be plugged in. Learning rate decay / scheduling can also be implemented a spart of this API.
70 | # 6. A metrics API which is used for assessing the performance of the model during training. A metric is the target to optimize during training, with specific metrics chosen for specific modeling objectives.
71 | # 7. A loss API that informs the model quantitatively how much it should try to minimize during training by providing a measure of error. Similar to metrics, specific loss functions are selected for specific modelung objectives.
72 | #
73 | # With the Functional API, our main workflow will follow the diagram below.
74 | #
75 | # :::{figure-md} Keras-fig
76 | #
77 | #
78 | # Keras Functional API diagram (from [https://miro.com/app/board/o9J_lhnKhVE=/](hhttps://miro.com/app/board/o9J_lhnKhVE=/)).
79 | # :::
80 |
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson2b_prep_data_ML_segmentation_35_1.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/jupyter_execute/docs/Lesson2b_prep_data_ML_segmentation_35_1.png
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson3_deeplearning_crop_segmentation_31_0.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/jupyter_execute/docs/Lesson3_deeplearning_crop_segmentation_31_0.png
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson3_deeplearning_crop_segmentation_45_0.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/jupyter_execute/docs/Lesson3_deeplearning_crop_segmentation_45_0.png
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson3_deeplearning_crop_segmentation_47_0.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/jupyter_execute/docs/Lesson3_deeplearning_crop_segmentation_47_0.png
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson3_deeplearning_crop_segmentation_49_0.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/jupyter_execute/docs/Lesson3_deeplearning_crop_segmentation_49_0.png
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson4_evaluation.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python
2 | # coding: utf-8
3 |
4 | # # Evaluating Semantic Segmentation Models
5 | # > A guide for understanding the performance of semantic segmentation for land use / land cover in satellite imagery.
6 |
7 | # After we have trained a model for segmenting images according to a set of classes it's time to evaluate how well it has performed. With any supervised machine learning model, we are typically interested in minimizing false positives or false negatives, with a preference for one or the other.
8 | #
9 | # Additionally, are also interested in making sure our model generalizes beyond the training dataset to new data it was not trained on. Finally, we'd like our model to make high confidence, correct predictions, and for higher confidence thresholds to not result in many more false negatives.
10 | #
11 | # For semantic segmentation, our evaluation unit is an individual pixel, which can be one of four categories:
12 | # * true positive: the pixel was classified correctly as a class of interest.
13 | # * true negative: the pixel was classified correctly as the background class.
14 | # * false positive: the pixel was incorrectly assigned a class of interest
15 | # * false negative: the pixel was incorrectly assigned the background class or a different class
16 | #
17 | # The most in depth and succinct summary we can produce is a confusion matrix. It summarrizes the counts of pixels that fall into each of these categories for each of our classes of interest and the background class.
18 | #
19 | #
20 | # 
21 | #
22 | # In this tutorial we will using data from a reference dataset hosted on Radiant Earth MLHub called ["A Fusion Dataset for Crop Type Classification in Germany"](https://mlhub.earth/data/dlr_fusion_competition_germany), and our U-Net predictions to compute a confusion matrix to assess our model performance.
23 | #
24 | # ## Specific concepts that will be covered
25 | # In the process, we will build practical experience and develop intuition around the following concepts:
26 | # * **[sci-kit learn](https://scikit-learn.org/stable/modules/generated/sklearn.metrics.confusion_matrix.html)** - we will use sci-kit learn to compute a confusion matrix and discuss how supplying different values to the `normalize `argument can help us interpret our data.
27 | # * **Metrics** - We will cover useful summary metrics that capture much of the information in the confusion matrix, including precision, recall, and F1 Score.
28 | #
29 | #
30 | # **Audience:** This post is geared towards intermediate users who are comfortable with basic machine learning concepts.
31 | #
32 | # **Time Estimated**: 60-120 min
33 | #
34 | #
35 |
36 | # ## Setup Notebook
37 |
38 | # In[ ]:
39 |
40 |
41 | # install required libraries
42 | get_ipython().system('pip install -q rasterio==1.2.10')
43 | get_ipython().system('pip install -q geopandas==0.10.2')
44 | get_ipython().system('pip install -q git+https://github.com/tensorflow/examples.git')
45 | get_ipython().system('pip install -q -U tfds-nightly')
46 | get_ipython().system('pip install -q focal-loss')
47 | get_ipython().system('pip install -q tensorflow-addons==0.8.3')
48 | #!pip install -q matplotlib==3.5 # UNCOMMENT if running on LOCAL
49 | get_ipython().system('pip install -q scikit-learn==1.0.1')
50 | get_ipython().system('pip install -q scikit-image==0.18.3')
51 |
52 |
53 | # In[2]:
54 |
55 |
56 | # import required libraries
57 | import os, glob, functools, fnmatch
58 | from zipfile import ZipFile
59 | from itertools import product
60 |
61 | import numpy as np
62 | import matplotlib.pyplot as plt
63 | import matplotlib as mpl
64 | mpl.rcParams['axes.grid'] = False
65 | mpl.rcParams['figure.figsize'] = (12,12)
66 | import matplotlib.image as mpimg
67 | import pandas as pd
68 | from PIL import Image
69 | import geopandas as gpd
70 | from IPython.display import clear_output
71 | from time import sleep
72 |
73 | import skimage.io as skio # lighter dependency than tensorflow for working with our tensors/arrays
74 | from sklearn.metrics import confusion_matrix, f1_score
75 |
76 |
77 | # #### Getting set up with the data
78 | #
79 | # ```{important}
80 | # The tiled imagery is available at the following path that is accessible with the google.colab `drive` module: `'/content/gdrive/My Drive/tf-eo-devseed/'`
81 | # ```
82 | #
83 | # We'll be working with the following folders and files in the `tf-eo-devseed` folder:
84 | # ```
85 | # tf-eo-devseed/
86 | # ├── stacks/
87 | # ├── stacks_brightened/
88 | # ├── indices/
89 | # ├── labels/
90 | # ├── background_list_train.txt
91 | # ├── train_list_clean.txt
92 | # └── lulc_classes.csv
93 | # ```
94 |
95 | # In[ ]:
96 |
97 |
98 | # set your root directory and tiled data folders
99 | if 'google.colab' in str(get_ipython()):
100 | # mount google drive
101 | from google.colab import drive
102 | drive.mount('/content/gdrive')
103 | root_dir = '/content/gdrive/My Drive/tf-eo-devseed/'
104 | workshop_dir = '/content/gdrive/My Drive/tf-eo-devseed-workshop'
105 | dirs = [root_dir, workshop_dir]
106 | for d in dirs:
107 | if not os.path.exists(d):
108 | os.makedirs(d)
109 | print('Running on Colab')
110 | else:
111 | root_dir = os.path.abspath("./data/tf-eo-devseed")
112 | workshop_dir = os.path.abspath('./tf-eo-devseed-workshop')
113 | print(f'Not running on Colab, data needs to be downloaded locally at {os.path.abspath(root_dir)}')
114 |
115 |
116 | # In[ ]:
117 |
118 |
119 | # go to root directory
120 | get_ipython().run_line_magic('cd', '$root_dir')
121 |
122 |
123 | # ### Check out the labels
124 | # Class names and identifiers extracted from the documentation provided here: https://radiantearth.blob.core.windows.net/mlhub/esa-food-security-challenge/Crops_GT_Brandenburg_Doc.pdf
125 | #
126 | # We'll use these labels to label our confusion matrix and table with class specific metrics.
127 |
128 | # In[7]:
129 |
130 |
131 | # Read the classes
132 |
133 | data = {'class_names': ['Background', 'Wheat', 'Rye', 'Barley', 'Oats', 'Corn', 'Oil Seeds', 'Root Crops', 'Meadows', 'Forage Crops'],
134 | 'class_ids': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
135 | }
136 |
137 | classes = pd.DataFrame(data)
138 | print(classes)
139 |
140 |
141 | # ### Getting image, prediction, and label filenames
142 | #
143 | # Before creating our confusion matrix, we need to associate labels, images, and prediction files together so that we know what to compare. We'll use our standard python for this.
144 |
145 | # ### Evaluate Model
146 | #
147 | # Compute confusion matrix from all predicted images and their ground truth label masks.
148 |
149 | # First we need to read in our prediction masks that we saved out in the last notebook. To do this, we can use scikit image, and then we can use scikit learn to compute our confusion matrix. No tensorflow needed for this part!
150 |
151 | # In[10]:
152 |
153 |
154 | path_df = pd.read_csv(os.path.join(workshop_dir, "test_file_paths.csv"))
155 |
156 |
157 | # In[ ]:
158 |
159 |
160 | pd.set_option('display.max_colwidth', None)
161 | path_df
162 |
163 |
164 | # In[12]:
165 |
166 |
167 | # reading in preds
168 | label_arr_lst = path_df["label_names"].apply(skio.imread)
169 | pred_arr_lst = path_df["pred_names"].apply(skio.imread)
170 |
171 |
172 | # A few of our labels have an image dimension that doesn't match the prediction dimension! It's possible this image was corrupted. We can skip it and the corresponding prediction when computing our metrics.
173 |
174 | # In[ ]:
175 |
176 |
177 | pred_arr_lst_valid = []
178 | label_arr_lst_valid = []
179 | for i in range(0, len(pred_arr_lst)):
180 | if pred_arr_lst[i].shape != label_arr_lst[i].shape:
181 |
182 | print(f"The {i}th label has an incorrect dimension, skipping.")
183 | print(pred_arr_lst[i])
184 | print(label_arr_lst[i])
185 | print(pred_arr_lst[i].shape)
186 | print(label_arr_lst[i].shape)
187 |
188 | else:
189 | pred_arr_lst_valid.append(pred_arr_lst[i])
190 | label_arr_lst_valid.append(label_arr_lst[i])
191 |
192 |
193 | # With our predictions and labels in lists of tiled arrays, we can then flatten these so that we instead have lists of pixels for predictions and labels. This is the format expected by scikit-learn's `confusion_matrix` function.
194 |
195 | # In[14]:
196 |
197 |
198 | # flatten our tensors and use scikit-learn to create a confusion matrix
199 | flat_preds = np.concatenate(pred_arr_lst_valid).flatten()
200 | flat_truth = np.concatenate(label_arr_lst_valid).flatten()
201 | OUTPUT_CHANNELS = 10
202 | cm = confusion_matrix(flat_truth, flat_preds, labels=list(range(OUTPUT_CHANNELS)))
203 |
204 |
205 | # Finally, we can plot the confusion matrix. We can either use the built-in method from scikit-learn...
206 |
207 | # In[ ]:
208 |
209 |
210 | from sklearn.metrics import ConfusionMatrixDisplay
211 |
212 | ConfusionMatrixDisplay.from_predictions(flat_truth, flat_preds, normalize='true')
213 |
214 |
215 | # ... or matplotlib, which allows us to more easily customize all aspects of our plot.
216 |
217 | # In[ ]:
218 |
219 |
220 | classes = [0,1,2,3,4,5,6,7,8,9]
221 |
222 | get_ipython().run_line_magic('matplotlib', 'inline')
223 | cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
224 | fig, ax = plt.subplots(figsize=(10, 10))
225 | im = ax.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)
226 | ax.figure.colorbar(im, ax=ax)
227 | # We want to show all ticks...
228 | ax.set(xticks=np.arange(cm.shape[1]),
229 | yticks=np.arange(cm.shape[0]),
230 | # ... and label them with the respective list entries
231 | xticklabels=list(range(OUTPUT_CHANNELS)), yticklabels=list(range(OUTPUT_CHANNELS)),
232 | title='Normalized Confusion Matrix',
233 | ylabel='True label',
234 | xlabel='Predicted label')
235 |
236 | # Rotate the tick labels and set their alignment.
237 | plt.setp(ax.get_xticklabels(), rotation=45, ha="right",
238 | rotation_mode="anchor")
239 |
240 | # Loop over data dimensions and create text annotations.
241 | fmt = '.2f' #'d' # if normalize else 'd'
242 | thresh = cm.max() / 2.
243 | for i in range(cm.shape[0]):
244 | for j in range(cm.shape[1]):
245 | ax.text(j, i, format(cm[i, j], fmt),
246 | ha="center", va="center",
247 | color="white" if cm[i, j] > thresh else "black")
248 | fig.tight_layout(pad=2.0, h_pad=2.0, w_pad=2.0)
249 | ax.set_ylim(len(classes)-0.5, -0.5)
250 |
251 |
252 | # Now let's compute the f1 score
253 | #
254 | # F1 = 2 * (precision * recall) / (precision + recall)
255 | #
256 |
257 | # You can view the documentation for a python function in a jupyter notebook with "??". scikit-learn docs usually come with detaile descriptions of each argument and examples of usage.
258 |
259 | # In[ ]:
260 |
261 |
262 | get_ipython().run_line_magic('pinfo2', 'f1_score')
263 |
264 |
265 | # In[ ]:
266 |
267 |
268 | # compute f1 score
269 | f1_score(flat_truth, flat_preds, average='macro')
270 |
271 |
272 | # In[ ]:
273 |
274 |
275 |
276 |
277 |
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson4_evaluation_21_1.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/jupyter_execute/docs/Lesson4_evaluation_21_1.png
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/Lesson4_evaluation_23_1.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/_build/jupyter_execute/docs/Lesson4_evaluation_23_1.png
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/appendix.ipynb:
--------------------------------------------------------------------------------
1 | {
2 | "cells": [
3 | {
4 | "cell_type": "markdown",
5 | "metadata": {},
6 | "source": [
7 | "# Appendix"
8 | ]
9 | }
10 | ],
11 | "metadata": {
12 | "kernelspec": {
13 | "display_name": "Python 3",
14 | "language": "python",
15 | "name": "python3"
16 | },
17 | "language_info": {
18 | "codemirror_mode": {
19 | "name": "ipython",
20 | "version": 3
21 | },
22 | "file_extension": ".py",
23 | "mimetype": "text/x-python",
24 | "name": "python",
25 | "nbconvert_exporter": "python",
26 | "pygments_lexer": "ipython3",
27 | "version": "3.8.11"
28 | }
29 | },
30 | "nbformat": 4,
31 | "nbformat_minor": 4
32 | }
--------------------------------------------------------------------------------
/ds_book/_build/jupyter_execute/docs/appendix.py:
--------------------------------------------------------------------------------
1 | #!/usr/bin/env python
2 | # coding: utf-8
3 |
4 | # # Appendix
5 |
--------------------------------------------------------------------------------
/ds_book/_config.yml:
--------------------------------------------------------------------------------
1 | title: "Deep learning with TensorFlow"
2 | author: "Development Seed"
3 | email: "lilly@developmentseed.org"
4 | release: "1"
5 | logo: 'ds.png'
6 |
7 | # Short description about the book
8 | description: >-
9 | Notebook tutorials demonstrating advanced techniques for use of deep learning with TensorFlow and earth observation data.
10 |
11 | execute:
12 | # execute_notebooks : auto
13 | execute_notebooks : off
14 | timeout: -1
15 |
16 | # HTML-specific settings
17 | html:
18 | favicon : "ds.png"
19 | home_page_in_navbar : false
20 | use_edit_page_button : true
21 | use_repository_button: true
22 | use_issues_button : true
23 |
24 | # Interact link settings
25 | notebook_interface : "notebook"
26 |
27 | repository:
28 | url : "https://github.com/developmentseed/tensorflow-eo-training/"
29 | branch : main
30 | path_to_book : ds_book
31 |
32 | # Launch button settings
33 | #launch_buttons:
34 | # binder : false
35 | # binderhub : false
36 | # binderhub_url : off
37 | # jupyterhub : false
38 | # thebe : false
39 | # colab : false
40 |
41 | # Launch button settings
42 | launch_buttons:
43 | # notebook_interface: classic #jupyterlab
44 | binderhub_url: https://mybinder.org
45 | colab_url: https://colab.research.google.com
46 |
47 |
48 | # LaTeX settings
49 | bibtex_bibfiles:
50 | - _bibliography/references.bib
51 |
52 | latex:
53 | latex_engine : "pdflatex"
54 | use_jupyterbook_latex : true # use jupyterbook-latex for pdf builds as default
55 | latex_documents:
56 | targetname: ds_book.tex
57 |
58 | sphinx:
59 | config:
60 | html_show_copyright: false
61 | latex_toplevel_sectioning: 'section'
62 | # latex_show_urls: 'footnote'
63 | latex_elements.papersize: a4paper
64 | latex_elements.pointsize: 12pt
65 |
66 |
67 | #google_analytics:
68 | # mytrackingcode: UA-205698170-1
69 |
--------------------------------------------------------------------------------
/ds_book/_toc.yml:
--------------------------------------------------------------------------------
1 | format: jb-book
2 | root: docs/index
3 | chapters:
4 | - file: docs/Lesson1a_Intro_ML_NN_DL
5 | - file: docs/Lesson1b_Intro_TensorFlow_Keras
6 | - file: docs/Lesson2a_get_planet_NICFI
7 | - file: docs/Lesson2b_prep_data_ML_segmentation
8 | - file: docs/Lesson3_deeplearning_crop_segmentation
9 | - file: docs/Lesson4_evaluation
10 | - file: docs/Lesson5_dealing_with_limited_data
11 | - file: docs/appendix
12 |
--------------------------------------------------------------------------------
/ds_book/covers/.DS_Store:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/covers/.DS_Store
--------------------------------------------------------------------------------
/ds_book/docs/.DS_Store:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/.DS_Store
--------------------------------------------------------------------------------
/ds_book/docs/Lesson1b_Intro_TensorFlow_Keras.ipynb:
--------------------------------------------------------------------------------
1 | {
2 | "cells": [
3 | {
4 | "cell_type": "markdown",
5 | "metadata": {},
6 | "source": [
7 | "# Introduction to TensorFlow and Keras"
8 | ]
9 | },
10 | {
11 | "cell_type": "markdown",
12 | "metadata": {},
13 | "source": [
14 | "## Objectives"
15 | ]
16 | },
17 | {
18 | "cell_type": "markdown",
19 | "metadata": {},
20 | "source": [
21 | "The goal of this notebook is to teach some basics of the TensorFlow framework and the Keras API. "
22 | ]
23 | },
24 | {
25 | "cell_type": "markdown",
26 | "metadata": {},
27 | "source": [
28 | "### What is TensorFlow?\n",
29 | "\n",
30 | "[TensorFlow](https://www.tensorflow.org/guide]) is an open-source framework developed in late 2015 by Google for building various machine learning and deep learning models. TensorFlow is free and open-source, thanks to the Apache Open Source license.\n",
31 | "\n",
32 | "The main objective of using TensorFlow is to reduce the complexity of implementing computations on large numerical data sets. In practice, these large computations can manifest as training and inference with machine learning or deep learning models.\n",
33 | "\n",
34 | "TensorFlow was designed to operate with multiple CPUs or GPUs, as well as a growing number of mobile operating systems. The framework includes wrappers in Python, C++, and Java.\n",
35 | "\n",
36 | "#### How does it work?\n",
37 | "TensorFlow accepts inputs as a multi-dimensional array called a Tensor, which allows the programmer to create dataflow graphs and structures specifying how data travels through. The framework is designed to support creation of a flowchart of operations to be applied to input Tensors, which travel in one direction and out the other.\n",
38 | "\n",
39 | "#### TensorFlow's structure\n",
40 | "There are three main components to TensorFlow's structure.\n",
41 | "\n",
42 | "1. preprocessing the data\n",
43 | "2. building the model\n",
44 | "3. training and estimating the model\n",
45 | "\n",
46 | "The name Tensorflow derives from the way in which the framework receives input in the form of a multi-dimensional array, i.e. the tensors. These tensors travel sequentially through the specified flowchart of the operations, entering at one end and culminating as output at the other end. \n",
47 | "\n",
48 | "#### What are the key TensorFlow components?\n",
49 | "**Tensor**\n",
50 | "\n",
51 | "A tensor is an n-dimensional vector or matrix. In theory, a tensor may represent any form of data. The values belonging to a tensor all share the same data type and often the same shape / dimensionality. A tensor can describe the input data and the output of a calculation. The main difference between a tensor and a conventional n-dimensional array is that tensors are immutable.\n",
52 | "\n",
53 | "\n",
54 | "**Graphs**\n",
55 | "\n",
56 | "TensorFlow uses a graph framework. Graphs collect and summarize all of the calculations and offer several benefits:\n",
57 | "\n",
58 | "1. They are designed to work on CPUs or GPUs, as well as on mobile devices.\n",
59 | "2. Graphs are portable which enables the computations to be saved for immediate or later usage. Otherwise stated, the graph can be frozen and run at a later time.\n",
60 | "3. Graph calculations are executed by linking tensors together.\n",
61 | "4. For each tensor, there is a node and an edge. The node carries out the mathematical process and produces endpoint outputs. The input/output connections are represented by the edges. \n",
62 | "5. All nodes are linked together, so the graph itself is a depiction of the operations and relationships that exist between the nodes.\n",
63 | "\n",
64 | "In TensorFlow, all operations are carried out within a graph, which in effect is a series of computations that happen in order. Each individual operation is referred to as an op node. \n",
65 | "\n",
66 | ":::{figure-md} TFgraph-fig\n",
67 | "\n",
68 | "\n",
69 | "TensorFlow graph example (from [https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4](https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4)).\n",
70 | ":::\n",
71 | "\n",
72 | "\n",
73 | "#### Why do so many people like TensorFlow?\n",
74 | "TensorFlow is intentionally user-friendly, with helpful plugins to visualize model training and a useful software debugging tool. As well, TensorFlow is highly scalable, with easy deployment on both CPUs and GPUs."
75 | ]
76 | },
77 | {
78 | "cell_type": "markdown",
79 | "metadata": {},
80 | "source": [
81 | "### What is Keras?\n",
82 | "\n",
83 | "[Keras](https://keras.io/about/) is an API built on Python which reduces the cognitive load associated with programming models through human readability and simple and consistent structures.\n",
84 | "\n",
85 | "Keras is what some might call a wrapper for TensorFlow. It is intended for rapid experimentation.\n",
86 | "\n",
87 | "Tha main components of Keras include:\n",
88 | "1. A models API, which enables one to construct a model with varying levels of complexity depending on use case. We will use the [Functional API](https://keras.io/guides/functional_api/) subclass.\n",
89 | "2. A layers API, which allows one to define the tensor in/tensor out computation functions.\n",
90 | "3. A callback API, which enables one to program specific actions to occur during training, such as log training metrics, visualize interim/internal states and statistics of the model during training, and perform early stopping when the model converges.\n",
91 | "4. A data preprocessing API, which offers support for prepping raw data from disk to model ready Tensor format.\n",
92 | "5. An optimizer API where all of the state of the art optimizers can be plugged in. Learning rate decay / scheduling can also be implemented as part of this API.\n",
93 | "6. A metrics API which is used for assessing the performance of the model during training. A metric is the target to optimize during training, with specific metrics chosen for specific modeling objectives.\n",
94 | "7. A loss API that informs the model quantitatively how much it should try to minimize during training by providing a measure of error. Similar to metrics, specific loss functions are selected for specific modeling objectives.\n",
95 | "\n",
96 | "With the Functional API, our main workflow will follow the diagram below.\n",
97 | "\n",
98 | ":::{figure-md} Keras-fig\n",
99 | "\n",
100 | "\n",
101 | "Keras Functional API diagram (from [https://miro.com/app/board/o9J_lhnKhVE=/](hhttps://miro.com/app/board/o9J_lhnKhVE=/)).\n",
102 | ":::"
103 | ]
104 | }
105 | ],
106 | "metadata": {
107 | "kernelspec": {
108 | "display_name": "Python 3",
109 | "language": "python",
110 | "name": "python3"
111 | },
112 | "language_info": {
113 | "codemirror_mode": {
114 | "name": "ipython",
115 | "version": 3
116 | },
117 | "file_extension": ".py",
118 | "mimetype": "text/x-python",
119 | "name": "python",
120 | "nbconvert_exporter": "python",
121 | "pygments_lexer": "ipython3",
122 | "version": "3.8.11"
123 | }
124 | },
125 | "nbformat": 4,
126 | "nbformat_minor": 2
127 | }
128 |
--------------------------------------------------------------------------------
/ds_book/docs/Lesson1b_annex.ipynb:
--------------------------------------------------------------------------------
1 | {
2 | "cells": [
3 | {
4 | "cell_type": "markdown",
5 | "metadata": {},
6 | "source": [
7 | "# Introduction to TensorFlow and Keras"
8 | ]
9 | },
10 | {
11 | "cell_type": "markdown",
12 | "metadata": {},
13 | "source": [
14 | "## Objectives"
15 | ]
16 | },
17 | {
18 | "cell_type": "markdown",
19 | "metadata": {},
20 | "source": [
21 | "The goal of this notebook is to teach some basics of the TensorFlow framework and the Keras API."
22 | ]
23 | },
24 | {
25 | "cell_type": "markdown",
26 | "metadata": {},
27 | "source": [
28 | "### What is TensorFlow?\n",
29 | "\n",
30 | "[TensorFlow](https://www.tensorflow.org/guide]) is an open-source framework developed by Google for building various machine learning and deep learning models. Although originally released in late 2015, the first stable version arrived in 2017. TensorFlow is free and open-source, thanks to the Apache Open Source license.\n",
31 | "\n",
32 | "The main objective of using TensorFlow is to reduce the complexity of implementing computations on large numerical data sets. In practice, these large computations can manifest as training and inference with machine learning or deep learning models.\n",
33 | "\n",
34 | "TensorFlow was designed to operate with multiple CPUs or GPUs, as well as a growing number of mobile operating systems. The framework includes wrappers in Python, C++, and Java.\n",
35 | "\n",
36 | "#### How does it work?\n",
37 | "TensorFlow accepts inputs as a multi-dimensional array called a Tensor, which allows the programmer to create dataflow graphs and structures specifying how data travels through. The framework is designed to support creation of a flowchart of operations to be applied to input Tensors, which travel in one direction and out the other.\n",
38 | "\n",
39 | "#### TensorFlow’s structure\n",
40 | "There are three main components to TensorFlow's structure.\n",
41 | "\n",
42 | "1. preprocessing the data\n",
43 | "2. building the model\n",
44 | "3. training and estimating the model\n",
45 | "\n",
46 | "The name TensorFlow derives from the way in which the framework receives input in the form of a multi-dimensional array, i.e. the tensors. These tensors travel sequentially through the specified flowchart of the operations, entering at one end and culminate as output at the other end. \n",
47 | "\n",
48 | "#### What are TensorFlow components?\n",
49 | "**Tensor**\n",
50 | "\n",
51 | "Tensors, the basic unit of data in this framework, are involved in every computation of TensorFlow. A tensor is an n-dimensional vector or matrix. In theory, a tensor may represent any form of data. The values belonging to a tensor all share the same data type and often the same shape / dimensionality. A tensor can describe the input data and the output of a calculation. \n",
52 | "\n",
53 | "In TensorFlow, all operations are carried out within a graph, which in effect is a series of computations that happen in order. Each indivisual operation is referred to as an op node. \n",
54 | "\n",
55 | "**Graphs**\n",
56 | "\n",
57 | "TensorFlow uses a graph framework. Graphs collect and summarize all of the calculations and offer several benefits:\n",
58 | "\n",
59 | "1. They are designed to work on CPUs or GPUs, as well as on mobile devices.\n",
60 | "2. Graphs are portable which enables the computations to be saved for immediate or later usage. Otherwsie stated, the graph can be frozen and run at a later time.\n",
61 | "3. Graph calculations are executed by linking tensors together.\n",
62 | "4. For each tensor, there is a node and an edge. The node carries out the mathematical process and produces endpoint outputs. The input/output connections are represented by the edges.\n",
63 | "5. All nodes are linked together, so the graph itself is a depiction of the operations and relationships that exist between the nodes.\n",
64 | "\n",
65 | ":::{figure-md} TFgraph-fig\n",
66 | "\n",
67 | "\n",
68 | "TensorFlow graph example (from [https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4](https://medium.com/the-artificial-impostor/notes-understanding-tensorflow-part-1-5f0ebb253ad4)).\n",
69 | ":::\n",
70 | "\n",
71 | "#### Why do so many people like TensorFlow?\n",
72 | "TensorFlow is intentionally user-friendly, with helpful plugins to visualize model training and a useful software debugging tool. As well, TensorFlow is highly scalable, with easy deployment on both CPUs and GPUs."
73 | ]
74 | },
75 | {
76 | "cell_type": "markdown",
77 | "metadata": {},
78 | "source": [
79 | "### What is Keras?\n",
80 | "\n",
81 | "[Keras](https://keras.io/about/) is an API built on Python, with human readability at the forefront of its design. With simple and consistent structures and methods extensible across many machine learning and deep learning applications, Keras reduces the cognitive load associated with programming models. Furthermore, the API seeks to minimize the need for programmer interaction by abstracting many complexities into easily callable functions. Lastly, Keras features clear & actionable error messaging, complemented by comprehensive and digestible documentation and developer guides.\n",
82 | "\n",
83 | "Keras is what some might call a wrapper for TensorFlow. By that, one means to say that Keras simplifies a programmer's interaction with TensorFlow through refinement of key methods and constructs.\n",
84 | "\n",
85 | "Importantly, Keras is intended for rapid experimentation.\n",
86 | "\n",
87 | "Tha main components of Keras include:\n",
88 | "1. A models API, which enables one to construct a model with varying levels of complexity depending on use case. We will use the [Functional API](https://keras.io/guides/functional_api/) subclass.\n",
89 | "2. A layers API, which allows one to define the tensor in/tnesor out computation functions.\n",
90 | "3. A callback API, which enables one to program specific actions to occur during training, such as log training metrics, visualize interim/internal states and statistics of the model during training, and perform early stopping when the model converges.\n",
91 | "4. A data preprocessing API, which offers support for prepping raw data from disk to model ready Tensor format.\n",
92 | "5. An optimizer API where all of the state of the art optimizers can be plugged in. Learning rate decay / scheduling can also be implemented a spart of this API.\n",
93 | "6. A metrics API which is used for assessing the performance of the model during training. A metric is the target to optimize during training, with specific metrics chosen for specific modeling objectives.\n",
94 | "7. A loss API that informs the model quantitatively how much it should try to minimize during training by providing a measure of error. Similar to metrics, specific loss functions are selected for specific modelung objectives.\n",
95 | "\n",
96 | "With the Functional API, our main workflow will follow the diagram below.\n",
97 | "\n",
98 | ":::{figure-md} Keras-fig\n",
99 | "\n",
100 | "\n",
101 | "Keras Functional API diagram (from [https://miro.com/app/board/o9J_lhnKhVE=/](hhttps://miro.com/app/board/o9J_lhnKhVE=/)).\n",
102 | ":::"
103 | ]
104 | }
105 | ],
106 | "metadata": {
107 | "kernelspec": {
108 | "display_name": "Python 3",
109 | "language": "python",
110 | "name": "python3"
111 | },
112 | "language_info": {
113 | "codemirror_mode": {
114 | "name": "ipython",
115 | "version": 3
116 | },
117 | "file_extension": ".py",
118 | "mimetype": "text/x-python",
119 | "name": "python",
120 | "nbconvert_exporter": "python",
121 | "pygments_lexer": "ipython3",
122 | "version": "3.8.11"
123 | }
124 | },
125 | "nbformat": 4,
126 | "nbformat_minor": 2
127 | }
128 |
--------------------------------------------------------------------------------
/ds_book/docs/appendix.ipynb:
--------------------------------------------------------------------------------
1 | {
2 | "cells": [
3 | {
4 | "cell_type": "markdown",
5 | "metadata": {},
6 | "source": [
7 | "# Appendix"
8 | ]
9 | }
10 | ],
11 | "metadata": {
12 | "kernelspec": {
13 | "display_name": "Python 3",
14 | "language": "python",
15 | "name": "python3"
16 | },
17 | "language_info": {
18 | "codemirror_mode": {
19 | "name": "ipython",
20 | "version": 3
21 | },
22 | "file_extension": ".py",
23 | "mimetype": "text/x-python",
24 | "name": "python",
25 | "nbconvert_exporter": "python",
26 | "pygments_lexer": "ipython3",
27 | "version": "3.8.11"
28 | }
29 | },
30 | "nbformat": 4,
31 | "nbformat_minor": 4
32 | }
33 |
--------------------------------------------------------------------------------
/ds_book/docs/images/.DS_Store:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/.DS_Store
--------------------------------------------------------------------------------
/ds_book/docs/images/Keras_functional_API.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/Keras_functional_API.jpg
--------------------------------------------------------------------------------
/ds_book/docs/images/Unet.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/Unet.png
--------------------------------------------------------------------------------
/ds_book/docs/images/Unet_mobilenetv2_arch_arch.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/Unet_mobilenetv2_arch_arch.png
--------------------------------------------------------------------------------
/ds_book/docs/images/cm.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/cm.png
--------------------------------------------------------------------------------
/ds_book/docs/images/epoch50_testimage.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/epoch50_testimage.png
--------------------------------------------------------------------------------
/ds_book/docs/images/epoch50_testimage1.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/epoch50_testimage1.png
--------------------------------------------------------------------------------
/ds_book/docs/images/kakarla2021.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/kakarla2021.png
--------------------------------------------------------------------------------
/ds_book/docs/images/loss_curve.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/loss_curve.png
--------------------------------------------------------------------------------
/ds_book/docs/images/lulc_labeling.gif:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/lulc_labeling.gif
--------------------------------------------------------------------------------
/ds_book/docs/images/marine_debris.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/marine_debris.png
--------------------------------------------------------------------------------
/ds_book/docs/images/neuralnet_basic.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/neuralnet_basic.png
--------------------------------------------------------------------------------
/ds_book/docs/images/neuron-structure.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/developmentseed/tensorflow-eo-training/80ad3f8f8d8031731adc19f36b4f8084f4c3e01b/ds_book/docs/images/neuron-structure.jpg
--------------------------------------------------------------------------------
/ds_book/docs/index.md:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 | # Deep Learning with TensorFlow: 
71 |
72 |
73 |