├── LICENSE ├── README.md ├── alexnet_data.py ├── alexnet_keras.py ├── alexnet_run.py ├── fig ├── 1.png ├── 2.png ├── 3.png ├── cor.png ├── cor_uncor.png └── uncor.png ├── generate_corr_data.py ├── generate_fake_data.py ├── lstm_keras.py └── requirements.txt /LICENSE: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "{}" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright {yyyy} {name of copyright owner} 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. 202 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # Using CNN on 2D Images of Time Series 2 | Because too often time series are fed as 1-D vectors Recurrent Neural Networks (Simple RNN, LSTM, GRU..). 3 | 4 |

5 | Will this time series go up or down in the next time frame?
6 | 7 |

8 | 9 |

10 | Which plot contains highly correlated time series?
11 | 12 |

13 | 14 | ## Possible advantages/drawbacks of such approach: 15 | 16 | ### Advantages 17 | - Almost no pre-processing. Feed raw pixels (be careful of the resolution of the image though)! 18 | - We can add several time series on the same plot or on a different plot and concatenate both images. 19 | - Conv Nets have the reputation of being more stable than Recurrent Neural Networks for many tasks (WaveNet is just one example). 20 | - No vanishing/exploding gradient! Even though, it's less true with LSTM. 21 | 22 | ### Drawbacks 23 | - Input is much bigger than feeding 1-D vectors. Actually it's very very sparse! 24 | - Training will be undoubtedly slower. 25 | - Sometimes it's also hard to train very big conv nets (VGG19 is such an example). 26 | 27 | 28 | ## Let's get started! 29 | 30 | ### Fake data generation 31 | ``` 32 | git clone https://github.com/philipperemy/tensorflow-cnn-time-series.git 33 | cd tensorflow-cnn-time-series/ 34 | sudo pip3 install -r requirements.txt 35 | python3 generate_data.py 36 | ``` 37 | 38 | ### Start the training of the CNN (AlexNet is used here) 39 | ``` 40 | python3 alexnet_run.py 41 | ``` 42 | 43 | ### Toy example: Binary classification of images of time series 44 | 45 | We consider the following binary classification problem of time series: 46 | - *UP*: If the time series went up in the next time frame. 47 | - *DOWN*: if the time series went down. 48 | 49 | Because it's impossible to classify pure random time series into two distinct classes, we expect a 50% accuracy on the testing set and the model to overfit on the training set. Here are some examples that we feed to the conv net: 50 | 51 |
52 |

53 |
54 | 55 |
56 |

57 |
58 | 59 | ### Keep in mind that LSTM is also good! 60 | ``` 61 | python3 lstm_keras.py # on correlation classification task 62 | [...] 63 | [test] loss= 0.021, acc= 100.00 64 | [test] loss= 0.004, acc= 100.00 65 | [test] loss= 0.004, acc= 100.00 66 | ``` 67 | -------------------------------------------------------------------------------- /alexnet_data.py: -------------------------------------------------------------------------------- 1 | import matplotlib 2 | 3 | matplotlib.use('Agg') 4 | 5 | from random import shuffle 6 | 7 | import errno 8 | import os 9 | from glob import glob 10 | import skimage.io 11 | import skimage.transform 12 | 13 | import matplotlib.pyplot as plt 14 | import numpy as np 15 | 16 | DATA_FOLDER = '/tmp/cnn-time-series/' 17 | 18 | 19 | def load_image(path): 20 | try: 21 | img = skimage.io.imread(path).astype(float) 22 | # TODO http://scikit-image.org/docs/dev/api/skimage.color.html rgb2gray 23 | # TODO cropping. 24 | img = skimage.transform.resize(img, (224, 224), mode='constant') 25 | except: 26 | return None 27 | if img is None: 28 | return None 29 | if len(img.shape) < 2: 30 | return None 31 | if len(img.shape) == 4: 32 | return None 33 | if len(img.shape) == 2: 34 | img = np.tile(img[:, :, None], 3) 35 | if img.shape[2] == 4: 36 | img = img[:, :, :3] 37 | if img.shape[2] > 4: 38 | return None 39 | 40 | img /= 255. 41 | return img 42 | 43 | 44 | def next_batch(x_y, index, batch_size): 45 | has_reset = False 46 | index *= batch_size 47 | updated_index = index % len(x_y) 48 | if updated_index + batch_size > len(x_y): 49 | updated_index = 0 50 | has_reset = True 51 | beg = updated_index 52 | end = updated_index + batch_size 53 | output = x_y[beg:end] 54 | x = np.array([e[0] for e in output]) 55 | y = np.array([e[1] for e in output]) 56 | return x, y, has_reset 57 | 58 | 59 | def read_dataset(folder, max_num_training_images, max_num_testing_images, class_mapper): 60 | training_inputs = read_set(folder, 'train', max_num_training_images, class_mapper) 61 | testing_inputs = read_set(folder, 'test', max_num_testing_images, class_mapper) 62 | return training_inputs, testing_inputs 63 | 64 | 65 | def read_set(folder, phase, max_num_of_images, class_mapper): 66 | images_folder = os.path.join(folder, phase) 67 | inputs = [] 68 | list_images = glob(images_folder + '/**/*.png') 69 | shuffle(list_images) 70 | for i, image_name in enumerate(list_images): 71 | if len(inputs) >= max_num_of_images: 72 | break 73 | class_name = image_name.split('/')[-2] 74 | if i % 100 == 0: 75 | print(i) 76 | inputs.append([load_image(image_name), class_mapper[class_name]]) # TODO make them 256x256 77 | return inputs 78 | 79 | 80 | def compute_mean_not_optimised(inputs): 81 | matrix_all_images = [] 82 | for image, label in inputs: 83 | matrix_all_images.append(image) 84 | return np.mean(np.array(matrix_all_images), axis=0) 85 | 86 | 87 | def compute_mean(inputs): 88 | image_mean = np.array(inputs[0][0]) 89 | image_mean.fill(0) 90 | for i, (image, label) in enumerate(inputs): 91 | image_mean += image 92 | if i % 100 == 0: 93 | print(i) 94 | return image_mean / len(inputs) 95 | 96 | 97 | def subtract_mean(inputs, mean_image): 98 | new_inputs = [] 99 | for image, label in inputs: 100 | new_inputs.append([image - mean_image, label]) 101 | return new_inputs 102 | 103 | 104 | def mkdir_p(path): 105 | try: 106 | os.makedirs(path) 107 | except OSError as exc: 108 | if exc.errno == errno.EEXIST and os.path.isdir(path): 109 | pass 110 | else: 111 | raise 112 | 113 | 114 | def generate_time_series(arr, filename): 115 | generate_multi_time_series([arr], filename) 116 | 117 | 118 | def generate_multi_time_series(arr_list, filename): 119 | fig = plt.figure() 120 | for arr in arr_list: 121 | plt.plot(arr) 122 | plt.savefig(filename) 123 | plt.close(fig) 124 | -------------------------------------------------------------------------------- /alexnet_keras.py: -------------------------------------------------------------------------------- 1 | import tensorflow as tf 2 | from keras.layers.convolutional import Conv2D, MaxPooling2D 3 | from keras.layers.core import Dense, Activation, Flatten 4 | 5 | 6 | # import tensorflow as tf 7 | # tf.python.control_flow_ops = tf # some hack to get tf running with Dropout 8 | 9 | # 224x224 10 | # https://gist.github.com/JBed/c2fb3ce8ed299f197eff 11 | def alex_net_keras(x, num_classes=2, keep_prob=0.5): 12 | x = Conv2D(92, kernel_size=(11, 11), strides=(4, 4), padding='same')(x) # conv 1 13 | # x = BatchNormalization()(x) 14 | x = Activation('relu')(x) 15 | # LRN is missing here - Caffe. 16 | x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2))(x) # pool 1 17 | 18 | x = Conv2D(256, kernel_size=(5, 5), padding='same')(x) # miss group and pad param # conv 2 19 | # x = BatchNormalization()(x) 20 | x = Activation('relu')(x) 21 | x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2))(x) # pool 2 22 | 23 | x = Conv2D(384, kernel_size=(3, 3), padding='same')(x) # conv 3 24 | # x = BatchNormalization()(x) 25 | x = Activation('relu')(x) 26 | # x = MaxPooling2D(pool_size=(3, 3))(x) 27 | 28 | x = Conv2D(384, kernel_size=(3, 3), padding='same')(x) # conv 4 29 | # x = BatchNormalization()(x) 30 | x = Activation('relu')(x) 31 | # x = MaxPooling2D(pool_size=(3, 3))(x) 32 | 33 | x = Conv2D(256, kernel_size=(3, 3), padding='same')(x) # conv 5 34 | # x = BatchNormalization()(x) 35 | x = Activation('relu')(x) 36 | 37 | x = Flatten()(x) 38 | x = Dense(4096, kernel_initializer='normal')(x) # fc6 39 | # dropout 0.5 40 | x = tf.nn.dropout(x, keep_prob=keep_prob) 41 | # x = BatchNormalization()(x) 42 | x = Activation('relu')(x) 43 | x = Dense(4096, kernel_initializer='normal')(x) # fc7 44 | # dropout 0.5 45 | x = tf.nn.dropout(x, keep_prob=keep_prob) 46 | # x = BatchNormalization()(x) 47 | x = Activation('relu')(x) 48 | x = Dense(num_classes)(x) 49 | # x = BatchNormalization()(x) 50 | # x = Activation('softmax')(x) 51 | return x 52 | -------------------------------------------------------------------------------- /alexnet_run.py: -------------------------------------------------------------------------------- 1 | import os 2 | 3 | import numpy as np 4 | import tensorflow as tf 5 | 6 | from alexnet_data import read_dataset, next_batch, compute_mean, subtract_mean 7 | from generate_fake_data import DATA_FOLDER 8 | 9 | if __name__ == '__main__': 10 | 11 | NUM_TRAINING_IMAGES = 1000 12 | NUM_TESTING_IMAGES = 1000 13 | 14 | names = os.listdir(os.path.join(DATA_FOLDER, 'train')) 15 | print(names) 16 | NUM_CLASSES = len(names) 17 | class_mapper = {names[0]: 0.0, names[1]: 1.0} 18 | print(class_mapper) 19 | BATCH_SIZE = 128 20 | HEIGHT = 224 21 | WIDTH = 224 22 | CHANNELS = 3 23 | LEARNING_RATE = 0.01 24 | data_percentage = 1 25 | num_training_images = data_percentage * NUM_TRAINING_IMAGES 26 | num_testing_images = data_percentage * NUM_TESTING_IMAGES 27 | 28 | print('read_dataset() start') 29 | training_inputs, testing_inputs = read_dataset(DATA_FOLDER, num_training_images, num_testing_images, class_mapper) 30 | print('read_dataset() done') 31 | print('compute_mean() start') 32 | mean_image = compute_mean(training_inputs) 33 | print('compute_mean() done') 34 | training_inputs = subtract_mean(training_inputs, mean_image) 35 | testing_inputs = subtract_mean(testing_inputs, mean_image) 36 | print(len(training_inputs), 'training inputs') 37 | print(len(testing_inputs), 'testing inputs') 38 | 39 | x = tf.placeholder(tf.float32, shape=[None, HEIGHT, WIDTH, CHANNELS]) 40 | y = tf.placeholder(tf.int64, [None]) 41 | keep_prob = tf.placeholder(tf.float32) 42 | 43 | from alexnet_keras import alex_net_keras 44 | 45 | logits = alex_net_keras(x, num_classes=len(names), keep_prob=keep_prob) 46 | 47 | cross_entropy = tf.reduce_mean(tf.nn.sparse_softmax_cross_entropy_with_logits(logits=logits, labels=y)) 48 | 49 | train_step = tf.train.GradientDescentOptimizer(LEARNING_RATE).minimize(cross_entropy) 50 | 51 | correct_prediction = tf.equal(tf.argmax(logits, 1), y) 52 | accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32)) 53 | 54 | sess = tf.Session() 55 | sess.run(tf.global_variables_initializer()) 56 | 57 | for i in range(int(1e9)): 58 | batch_xs, batch_ys, _ = next_batch(training_inputs, i, BATCH_SIZE) 59 | tr_loss, _ = sess.run([cross_entropy, train_step], 60 | feed_dict={x: batch_xs, y: batch_ys, keep_prob: 0.5}) 61 | print('[TRAINING] #batch = {0}, tr_loss = {1:.3f}'.format(i, tr_loss)) 62 | if i % 100 == 0: 63 | accuracy_list = [] 64 | j = 0 65 | while True: 66 | batch_xt, batch_yt, reset = next_batch(testing_inputs, j, BATCH_SIZE) 67 | if reset: 68 | break 69 | te_loss, te_acc = sess.run([cross_entropy, accuracy], 70 | feed_dict={x: batch_xt, y: batch_yt, keep_prob: 1.0}) 71 | print('[TESTING] #batch = {0}, te_loss = {1:.3f}, te_acc = {2:.3f}'.format(i, te_loss, te_acc)) 72 | accuracy_list.append(te_acc) 73 | j += 1 74 | print('[ALL] total batches = {0} total mean accuracy on testing set = {1:.2f}'.format(i, np.mean( 75 | accuracy_list))) 76 | -------------------------------------------------------------------------------- /fig/1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/philipperemy/tensorflow-cnn-time-series/d975e822b824315dfc6fa11a5af0450e42c6b0bf/fig/1.png -------------------------------------------------------------------------------- /fig/2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/philipperemy/tensorflow-cnn-time-series/d975e822b824315dfc6fa11a5af0450e42c6b0bf/fig/2.png -------------------------------------------------------------------------------- /fig/3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/philipperemy/tensorflow-cnn-time-series/d975e822b824315dfc6fa11a5af0450e42c6b0bf/fig/3.png -------------------------------------------------------------------------------- /fig/cor.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/philipperemy/tensorflow-cnn-time-series/d975e822b824315dfc6fa11a5af0450e42c6b0bf/fig/cor.png -------------------------------------------------------------------------------- /fig/cor_uncor.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/philipperemy/tensorflow-cnn-time-series/d975e822b824315dfc6fa11a5af0450e42c6b0bf/fig/cor_uncor.png -------------------------------------------------------------------------------- /fig/uncor.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/philipperemy/tensorflow-cnn-time-series/d975e822b824315dfc6fa11a5af0450e42c6b0bf/fig/uncor.png -------------------------------------------------------------------------------- /generate_corr_data.py: -------------------------------------------------------------------------------- 1 | import shutil 2 | 3 | import progressbar 4 | import scipy.stats as stats 5 | 6 | from alexnet_data import * 7 | 8 | 9 | def generate_two_correlated_time_series(size, rho): 10 | num_samples = size 11 | num_variables = 2 12 | cov = [[1.0, rho], [rho, 1.0]] 13 | 14 | L = np.linalg.cholesky(cov) 15 | 16 | uncorrelated = np.random.standard_normal((num_variables, num_samples)) 17 | correlated = np.dot(L, uncorrelated) 18 | x, y = correlated 19 | rho, p_val = stats.pearsonr(x, y) 20 | return x, y, rho 21 | 22 | 23 | def generate(): 24 | shutil.rmtree(DATA_FOLDER, ignore_errors=True) 25 | size = 1024 26 | total_num_images = 100000 27 | bar = progressbar.ProgressBar() 28 | for i in bar(range(total_num_images)): 29 | if i % 2 == 0: 30 | class_name = 'CORRELATED' 31 | rho = 0.8 32 | else: 33 | class_name = 'UNCORRELATED' 34 | rho = 0.0 35 | 36 | # more fun to consider non-stationary time series. 37 | # still the correlation holds almost surely. 38 | x_tr, y_tr, _ = generate_two_correlated_time_series(size, rho) 39 | x_tr = np.cumsum(x_tr) 40 | y_tr = np.cumsum(y_tr) 41 | 42 | x_te, y_te, _ = generate_two_correlated_time_series(size, rho) 43 | x_te = np.cumsum(x_te) 44 | y_te = np.cumsum(y_te) 45 | 46 | train_output_dir = os.path.join(DATA_FOLDER, 'train', class_name) 47 | mkdir_p(train_output_dir) 48 | generate_multi_time_series([x_tr, y_tr], os.path.join(train_output_dir, '{}.png'.format(i))) 49 | 50 | test_output_dir = os.path.join(DATA_FOLDER, 'test', class_name) 51 | mkdir_p(test_output_dir) 52 | generate_multi_time_series([x_te, y_te], os.path.join(test_output_dir, '{}.png'.format(i))) 53 | 54 | 55 | if __name__ == '__main__': 56 | generate() 57 | -------------------------------------------------------------------------------- /generate_fake_data.py: -------------------------------------------------------------------------------- 1 | from alexnet_data import * 2 | 3 | 4 | def generate(): 5 | for i in range(200): 6 | if i % 2 == 0: 7 | direction = 'UP' 8 | else: 9 | direction = 'DOWN' 10 | train_output_dir = os.path.join(DATA_FOLDER, 'train', direction) 11 | mkdir_p(train_output_dir) 12 | arr = np.cumsum(np.random.standard_normal(1024)) 13 | generate_time_series(arr, os.path.join(train_output_dir, 'img_{}.png'.format(i))) 14 | 15 | test_output_dir = os.path.join(DATA_FOLDER, 'test', direction) 16 | mkdir_p(test_output_dir) 17 | # arr = np.cumsum(np.random.standard_normal(1024)) 18 | generate_time_series(arr, os.path.join(test_output_dir, 'img_{}.png'.format(i))) 19 | 20 | 21 | if __name__ == '__main__': 22 | generate() 23 | -------------------------------------------------------------------------------- /lstm_keras.py: -------------------------------------------------------------------------------- 1 | import numpy as np 2 | from keras.layers import Dense 3 | from keras.layers import LSTM 4 | from keras.models import Sequential 5 | 6 | from generate_corr_data import generate_two_correlated_time_series 7 | 8 | max_len = 1024 9 | batch_size = 512 10 | 11 | 12 | def next_batch(): 13 | x = [] 14 | y = [] 15 | for i in range(batch_size // 2): 16 | x.append(generate_two_correlated_time_series(size=max_len, rho=0.8)[0:2]) 17 | y.append(1.0) 18 | 19 | x.append(generate_two_correlated_time_series(size=max_len, rho=0)[0:2]) 20 | y.append(0.0) 21 | return np.transpose(np.array(x), (0, 2, 1)), np.array(y) 22 | 23 | 24 | print('Build model...') 25 | model = Sequential() 26 | model.add(LSTM(128, dropout=0.2, recurrent_dropout=0.2, batch_input_shape=(batch_size, max_len, 2), 27 | return_sequences=True)) 28 | model.add(LSTM(128, dropout=0.2, recurrent_dropout=0.2)) 29 | model.add(Dense(128)) 30 | model.add(Dense(128)) 31 | model.add(Dense(1, activation='sigmoid')) 32 | print(model.summary()) 33 | 34 | # try using different optimizers and different optimizer configs 35 | model.compile(loss='binary_crossentropy', 36 | optimizer='adam', 37 | metrics=['accuracy']) 38 | 39 | print('Train...') 40 | while True: 41 | x_train, y_train = next_batch() 42 | train_loss, train_acc = model.train_on_batch(x_train, y_train) 43 | print('[train] loss= {0:.3f}, acc= {1:.2f}'.format(train_loss, train_acc * 100)) 44 | 45 | x_test, y_test = next_batch() 46 | test_loss, test_acc = model.test_on_batch(x_test, y_test) 47 | print('[test] loss= {0:.3f}, acc= {1:.2f}'.format(test_loss, test_acc * 100)) 48 | -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | tensorflow 2 | Keras 3 | matplotlib 4 | numpy 5 | scikit_image 6 | --------------------------------------------------------------------------------