├── README.md
├── Segundo_projeto_aula1.ipynb
├── Segundo_projeto_aula2.ipynb
├── Segundo_projeto_aula3.ipynb
├── Segundo_projeto_aula4.ipynb
└── Segundo_projeto_aula5.ipynb


/README.md:
--------------------------------------------------------------------------------
1 | # keras-tensorflow-2
2 | Repo para os arquivos do segundo curso de Keras na Alura
3 | 


--------------------------------------------------------------------------------
/Segundo_projeto_aula1.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |   "nbformat": 4,
  3 |   "nbformat_minor": 0,
  4 |   "metadata": {
  5 |     "colab": {
  6 |       "name": "Segundo_projeto_aula1.ipynb",
  7 |       "version": "0.3.2",
  8 |       "provenance": [],
  9 |       "collapsed_sections": []
 10 |     },
 11 |     "kernelspec": {
 12 |       "name": "python3",
 13 |       "display_name": "Python 3"
 14 |     }
 15 |   },
 16 |   "cells": [
 17 |     {
 18 |       "metadata": {
 19 |         "id": "sNc3NouWYrN7",
 20 |         "colab_type": "text"
 21 |       },
 22 |       "cell_type": "markdown",
 23 |       "source": [
 24 |         "Imports"
 25 |       ]
 26 |     },
 27 |     {
 28 |       "metadata": {
 29 |         "id": "4bQCahYjYdUB",
 30 |         "colab_type": "code",
 31 |         "colab": {}
 32 |       },
 33 |       "cell_type": "code",
 34 |       "source": [
 35 |         "import tensorflow\n",
 36 |         "from tensorflow import keras\n",
 37 |         "import matplotlib.pyplot as plt\n",
 38 |         "import numpy as np\n",
 39 |         "from tensorflow.keras.models import load_model"
 40 |       ],
 41 |       "execution_count": 0,
 42 |       "outputs": []
 43 |     },
 44 |     {
 45 |       "metadata": {
 46 |         "id": "f8uuzUjVYvkf",
 47 |         "colab_type": "text"
 48 |       },
 49 |       "cell_type": "markdown",
 50 |       "source": [
 51 |         "Carregando o dataset"
 52 |       ]
 53 |     },
 54 |     {
 55 |       "metadata": {
 56 |         "id": "QoIys3wMYirH",
 57 |         "colab_type": "code",
 58 |         "colab": {}
 59 |       },
 60 |       "cell_type": "code",
 61 |       "source": [
 62 |         "dataset = keras.datasets.fashion_mnist\n",
 63 |         "((imagens_treino, identificacoes_treino), (imagens_teste, identificacoes_teste)) = dataset.load_data()\n"
 64 |       ],
 65 |       "execution_count": 0,
 66 |       "outputs": []
 67 |     },
 68 |     {
 69 |       "metadata": {
 70 |         "id": "oFr0526ZY5E_",
 71 |         "colab_type": "text"
 72 |       },
 73 |       "cell_type": "markdown",
 74 |       "source": [
 75 |         "Exploração dos dados"
 76 |       ]
 77 |     },
 78 |     {
 79 |       "metadata": {
 80 |         "id": "fPIT36hpYl5A",
 81 |         "colab_type": "code",
 82 |         "colab": {
 83 |           "base_uri": "https://localhost:8080/",
 84 |           "height": 34
 85 |         },
 86 |         "outputId": "61176d4b-23cf-4b16-887b-9c6a5532ad9f"
 87 |       },
 88 |       "cell_type": "code",
 89 |       "source": [
 90 |         "len(imagens_treino)\n",
 91 |         "imagens_treino.shape\n",
 92 |         "imagens_teste.shape\n",
 93 |         "len(identificacoes_teste)\n",
 94 |         "identificacoes_treino.min()\n",
 95 |         "identificacoes_treino.max()"
 96 |       ],
 97 |       "execution_count": 3,
 98 |       "outputs": [
 99 |         {
100 |           "output_type": "execute_result",
101 |           "data": {
102 |             "text/plain": [
103 |               "9"
104 |             ]
105 |           },
106 |           "metadata": {
107 |             "tags": []
108 |           },
109 |           "execution_count": 3
110 |         }
111 |       ]
112 |     },
113 |     {
114 |       "metadata": {
115 |         "id": "yTWqT9DIY-iB",
116 |         "colab_type": "text"
117 |       },
118 |       "cell_type": "markdown",
119 |       "source": [
120 |         "Exibição dos dados"
121 |       ]
122 |     },
123 |     {
124 |       "metadata": {
125 |         "id": "bGESm49JVahh",
126 |         "colab_type": "code",
127 |         "colab": {
128 |           "base_uri": "https://localhost:8080/",
129 |           "height": 364
130 |         },
131 |         "outputId": "c23e4dd9-50c3-4863-8813-8aeb793c49cf"
132 |       },
133 |       "cell_type": "code",
134 |       "source": [
135 |         "total_de_classificacoes = 10\n",
136 |         "nomes_de_classificacoes = ['Camiseta', 'Calça', 'Pullover', \n",
137 |         "                           'Vestido', 'Casaco', 'Sandália', 'Camisa',\n",
138 |         "                           'Tênis', 'Bolsa', 'Bota']\n",
139 |         "\n",
140 |         "plt.imshow(imagens_treino[0])\n",
141 |         "plt.colorbar()"
142 |       ],
143 |       "execution_count": 4,
144 |       "outputs": [
145 |         {
146 |           "output_type": "execute_result",
147 |           "data": {
148 |             "text/plain": [
149 |               "<matplotlib.colorbar.Colorbar at 0x7f99d2827a20>"
150 |             ]
151 |           },
152 |           "metadata": {
153 |             "tags": []
154 |           },
155 |           "execution_count": 4
156 |         },
157 |         {
158 |           "output_type": "display_data",
159 |           "data": {
160 |             "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAFKCAYAAAAHY8iZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
161 |             "text/plain": [
162 |               "<matplotlib.figure.Figure at 0x7f99d28f0f60>"
163 |             ]
164 |           },
165 |           "metadata": {
166 |             "tags": []
167 |           }
168 |         }
169 |       ]
170 |     },
171 |     {
172 |       "metadata": {
173 |         "id": "uU6ZySpIcoc5",
174 |         "colab_type": "text"
175 |       },
176 |       "cell_type": "markdown",
177 |       "source": [
178 |         "Normalizando as imagens"
179 |       ]
180 |     },
181 |     {
182 |       "metadata": {
183 |         "id": "Ral_hdl9ulGG",
184 |         "colab_type": "code",
185 |         "colab": {}
186 |       },
187 |       "cell_type": "code",
188 |       "source": [
189 |         "imagens_treino = imagens_treino/float(255)"
190 |       ],
191 |       "execution_count": 0,
192 |       "outputs": []
193 |     },
194 |     {
195 |       "metadata": {
196 |         "id": "2WABnvOJdCNl",
197 |         "colab_type": "text"
198 |       },
199 |       "cell_type": "markdown",
200 |       "source": [
201 |         "Criando, compilando, treinando e normalizando o modelo"
202 |       ]
203 |     },
204 |     {
205 |       "metadata": {
206 |         "id": "uFedwlfFc0ii",
207 |         "colab_type": "code",
208 |         "colab": {
209 |           "base_uri": "https://localhost:8080/",
210 |           "height": 226
211 |         },
212 |         "outputId": "7772c07b-077d-43e8-a9cd-7f7055df320f"
213 |       },
214 |       "cell_type": "code",
215 |       "source": [
216 |         "modelo = keras.Sequential([ \n",
217 |         "    keras.layers.Flatten(input_shape=(28, 28)),\n",
218 |         "    keras.layers.Dense(256, activation=tensorflow.nn.relu),\n",
219 |         "    keras.layers.Dropout(0.2),\n",
220 |         "    keras.layers.Dense(10, activation=tensorflow.nn.softmax)\n",
221 |         "])\n",
222 |         "\n",
223 |         "modelo.compile(optimizer='adam', \n",
224 |         "               loss='sparse_categorical_crossentropy',\n",
225 |         "               metrics=['accuracy'])\n",
226 |         "\n",
227 |         "historico = modelo.fit(imagens_treino, identificacoes_treino, epochs=5, validation_split=0.2)"
228 |       ],
229 |       "execution_count": 6,
230 |       "outputs": [
231 |         {
232 |           "output_type": "stream",
233 |           "text": [
234 |             "Train on 48000 samples, validate on 12000 samples\n",
235 |             "Epoch 1/5\n",
236 |             "48000/48000 [==============================] - 9s 189us/step - loss: 0.5252 - acc: 0.8134 - val_loss: 0.4049 - val_acc: 0.8524\n",
237 |             "Epoch 2/5\n",
238 |             "48000/48000 [==============================] - 9s 183us/step - loss: 0.3959 - acc: 0.8569 - val_loss: 0.3694 - val_acc: 0.8693\n",
239 |             "Epoch 3/5\n",
240 |             "48000/48000 [==============================] - 9s 182us/step - loss: 0.3605 - acc: 0.8683 - val_loss: 0.3779 - val_acc: 0.8627\n",
241 |             "Epoch 4/5\n",
242 |             "48000/48000 [==============================] - 9s 183us/step - loss: 0.3377 - acc: 0.8757 - val_loss: 0.3443 - val_acc: 0.8755\n",
243 |             "Epoch 5/5\n",
244 |             "48000/48000 [==============================] - 9s 182us/step - loss: 0.3209 - acc: 0.8805 - val_loss: 0.3495 - val_acc: 0.8765\n"
245 |           ],
246 |           "name": "stdout"
247 |         }
248 |       ]
249 |     },
250 |     {
251 |       "metadata": {
252 |         "id": "weeyzt2QuDXS",
253 |         "colab_type": "text"
254 |       },
255 |       "cell_type": "markdown",
256 |       "source": [
257 |         "Sumário do modelo"
258 |       ]
259 |     },
260 |     {
261 |       "metadata": {
262 |         "colab_type": "code",
263 |         "outputId": "5bf36159-d992-43e3-9f62-a2950eeb7628",
264 |         "id": "QLG_CH6Lr2vp",
265 |         "colab": {
266 |           "base_uri": "https://localhost:8080/",
267 |           "height": 274
268 |         }
269 |       },
270 |       "cell_type": "code",
271 |       "source": [
272 |         "sumario_do_modelo = modelo.summary()"
273 |       ],
274 |       "execution_count": 38,
275 |       "outputs": [
276 |         {
277 |           "output_type": "stream",
278 |           "text": [
279 |             "_________________________________________________________________\n",
280 |             "Layer (type)                 Output Shape              Param #   \n",
281 |             "=================================================================\n",
282 |             "flatten (Flatten)            (None, 784)               0         \n",
283 |             "_________________________________________________________________\n",
284 |             "dense (Dense)                (None, 256)               200960    \n",
285 |             "_________________________________________________________________\n",
286 |             "dropout (Dropout)            (None, 256)               0         \n",
287 |             "_________________________________________________________________\n",
288 |             "dense_1 (Dense)              (None, 10)                2570      \n",
289 |             "=================================================================\n",
290 |             "Total params: 203,530\n",
291 |             "Trainable params: 203,530\n",
292 |             "Non-trainable params: 0\n",
293 |             "_________________________________________________________________\n"
294 |           ],
295 |           "name": "stdout"
296 |         }
297 |       ]
298 |     },
299 |     {
300 |       "metadata": {
301 |         "id": "GVuSQg42udch",
302 |         "colab_type": "text"
303 |       },
304 |       "cell_type": "markdown",
305 |       "source": [
306 |         "Pesos da primeira camada Dense"
307 |       ]
308 |     },
309 |     {
310 |       "metadata": {
311 |         "id": "YRzqJi5gr-ug",
312 |         "colab_type": "code",
313 |         "colab": {
314 |           "base_uri": "https://localhost:8080/",
315 |           "height": 1354
316 |         },
317 |         "outputId": "fb501c8a-69f6-49c1-ae48-176d4cf06592"
318 |       },
319 |       "cell_type": "code",
320 |       "source": [
321 |         "pesos_camada_dense = modelo.layers[1].get_weights()\n",
322 |         "pesos_camada_dense"
323 |       ],
324 |       "execution_count": 40,
325 |       "outputs": [
326 |         {
327 |           "output_type": "execute_result",
328 |           "data": {
329 |             "text/plain": [
330 |               "[array([[ 0.00524871, -0.11004062,  0.13758828, ..., -0.1579878 ,\n",
331 |               "         -0.00142127,  0.00939471],\n",
332 |               "        [ 0.03461614, -0.04760416,  0.1779013 , ..., -0.14755748,\n",
333 |               "         -0.10598074, -0.05744827],\n",
334 |               "        [-0.04720343,  0.01652463, -0.03128776, ..., -0.13856664,\n",
335 |               "          0.01710287, -0.05725543],\n",
336 |               "        ...,\n",
337 |               "        [ 0.03753441,  0.21444741,  0.01582495, ..., -0.2257942 ,\n",
338 |               "          0.2782449 , -0.08317742],\n",
339 |               "        [-0.0208508 ,  0.22376445,  0.00300794, ..., -0.17215742,\n",
340 |               "          0.04819855, -0.11253677],\n",
341 |               "        [ 0.02667857,  0.08448928,  0.23865685, ..., -0.05406902,\n",
342 |               "         -0.14307678,  0.01467239]], dtype=float32),\n",
343 |               " array([ 1.16563626e-01,  2.00383574e-01,  2.16542244e-01, -1.04050189e-01,\n",
344 |               "         2.22290292e-01,  2.03423440e-01,  1.78649172e-01,  1.04543686e-01,\n",
345 |               "         3.23054373e-01,  2.97243536e-01, -5.07429475e-03,  1.30689308e-01,\n",
346 |               "         3.12031865e-01,  3.39405507e-01,  1.46878883e-01,  2.59472758e-01,\n",
347 |               "        -1.03004798e-01,  3.85944843e-01,  3.57965261e-01, -7.44848475e-02,\n",
348 |               "         1.53556302e-01,  2.76009977e-01,  1.60143286e-01,  3.65746707e-01,\n",
349 |               "         3.35957766e-01, -4.07021679e-02,  3.53741157e-03,  1.49494678e-01,\n",
350 |               "        -1.18720653e-02,  2.86895782e-01,  1.71697333e-01,  2.70660341e-01,\n",
351 |               "        -1.90745622e-01, -1.26834903e-02, -3.37902158e-02,  1.94578290e-01,\n",
352 |               "         3.59597266e-01,  1.14150092e-01,  1.12235427e-01,  3.16841990e-01,\n",
353 |               "         9.54611879e-03, -1.95544466e-01,  2.29236320e-01,  1.89695671e-01,\n",
354 |               "         4.73570004e-02,  9.87034738e-02,  2.32090458e-01,  2.12860946e-02,\n",
355 |               "         6.61147758e-02, -4.24835980e-01,  3.15535158e-01,  3.21921289e-01,\n",
356 |               "         7.37581030e-02,  3.09438467e-01, -1.87978446e-01, -2.61927731e-02,\n",
357 |               "        -1.60711899e-01,  2.42159218e-01,  1.56536087e-01,  9.22217965e-03,\n",
358 |               "         1.82041883e-01,  9.23544243e-02,  1.35896012e-01,  2.39720523e-01,\n",
359 |               "        -8.12186971e-02,  1.23139266e-02, -3.39500867e-02,  1.22392870e-01,\n",
360 |               "         2.00472146e-01, -2.17560562e-03, -2.20949188e-01,  2.72897154e-01,\n",
361 |               "        -1.08623579e-02,  1.96204364e-01,  3.35872024e-01, -1.97613165e-01,\n",
362 |               "         2.76348889e-01,  1.17945582e-01,  1.88796625e-01,  2.31201984e-02,\n",
363 |               "        -1.41431298e-02,  3.27661216e-01,  2.72685587e-02,  2.58330017e-01,\n",
364 |               "         4.87888791e-02,  2.10950270e-01,  3.46322268e-01,  3.43774706e-01,\n",
365 |               "        -4.27812524e-02,  2.40779400e-01, -6.12564832e-02, -1.78985372e-01,\n",
366 |               "        -8.94419942e-03,  2.07404166e-01,  5.10557322e-03, -1.01498179e-02,\n",
367 |               "         1.30236566e-01,  2.50320852e-01,  2.00364500e-01,  6.64655566e-02,\n",
368 |               "        -1.01419687e-01,  2.57371038e-01, -2.33354792e-01,  3.01758826e-01,\n",
369 |               "         2.40178391e-01,  1.14074878e-01, -1.13115385e-02, -3.27479141e-03,\n",
370 |               "         2.65459508e-01, -1.20592214e-01,  3.37435186e-01, -7.84784369e-03,\n",
371 |               "         2.28633240e-01,  9.78518352e-02,  3.25709015e-01,  2.96632439e-01,\n",
372 |               "         4.44317120e-04,  1.59170657e-01,  4.48305488e-01,  1.08523689e-01,\n",
373 |               "         7.20590055e-02, -9.18169916e-02,  9.71861929e-02, -2.16058847e-02,\n",
374 |               "         5.89288995e-02,  1.77676991e-01, -3.08794882e-02,  2.08451599e-01,\n",
375 |               "         3.34183648e-02,  1.13854900e-01,  1.96839243e-01,  1.40792608e-01,\n",
376 |               "         5.55239134e-02,  1.56214744e-01,  7.90987816e-03,  3.63177717e-01,\n",
377 |               "        -1.56363193e-02,  3.68518591e-01,  9.49568376e-02, -2.08511874e-01,\n",
378 |               "         2.18438044e-01,  6.51510805e-02,  5.28218560e-02, -1.42783836e-01,\n",
379 |               "         1.68985445e-02,  2.90053546e-01,  6.10899851e-02, -1.64991133e-02,\n",
380 |               "         2.13183105e-01,  1.58428669e-01, -6.41118810e-02,  2.65895650e-02,\n",
381 |               "         1.05100922e-01,  1.29719332e-01,  1.71137169e-01,  4.18035805e-01,\n",
382 |               "        -1.75026402e-01, -1.79482233e-02,  1.49668813e-01,  2.35473365e-01,\n",
383 |               "        -2.79365368e-02,  9.41284597e-02, -1.01978900e-02,  1.24944352e-01,\n",
384 |               "         1.08707353e-01, -7.94230867e-03, -3.24166775e-01, -3.66591185e-01,\n",
385 |               "         2.68817842e-01,  2.87170887e-01,  1.16077222e-01, -1.23165967e-02,\n",
386 |               "         1.69344917e-01,  2.15018451e-01, -5.96200190e-02,  1.81544319e-01,\n",
387 |               "        -3.78576550e-03,  2.07346842e-01,  2.60249138e-01, -1.85766026e-01,\n",
388 |               "         1.19951367e-01,  6.72665611e-02,  2.02264652e-01,  1.97026134e-01,\n",
389 |               "         1.78503782e-01,  2.66332686e-01, -2.97352634e-02,  1.04686908e-01,\n",
390 |               "         3.03999603e-01, -1.34745538e-02, -1.50592193e-01, -2.33385824e-02,\n",
391 |               "        -2.67395973e-02,  3.34146231e-01,  4.63848084e-01,  9.08255056e-02,\n",
392 |               "        -1.28224254e-01,  1.70217261e-01, -4.45119590e-02, -2.88670324e-02,\n",
393 |               "        -2.16365922e-02,  3.31768245e-01,  4.08323035e-02,  1.36680275e-01,\n",
394 |               "         8.81125554e-02, -6.27287403e-02,  2.18742967e-01,  1.92244291e-01,\n",
395 |               "         2.81801503e-02,  1.92194059e-01,  5.07624567e-01, -2.77130216e-01,\n",
396 |               "         5.13272882e-02,  2.02881262e-01,  2.07814813e-01,  2.32015803e-01,\n",
397 |               "         8.77994075e-02,  3.72730494e-01, -8.68224353e-03,  8.90769064e-02,\n",
398 |               "         2.78868377e-01, -1.33509599e-02, -1.02290526e-01,  9.12050828e-02,\n",
399 |               "         9.56115499e-02, -1.45345896e-01,  9.74815190e-02,  3.14872265e-01,\n",
400 |               "         4.59881350e-02,  7.03799501e-02,  1.48478866e-01, -2.41981402e-01,\n",
401 |               "         2.21961141e-02, -3.69806541e-03, -9.29736793e-02,  2.39182755e-01,\n",
402 |               "         2.17786700e-01,  1.10879233e-02,  2.37261593e-01,  5.18026769e-01,\n",
403 |               "        -1.78707734e-01, -2.14400850e-02,  1.63810670e-01, -2.38751739e-01,\n",
404 |               "         2.65253652e-02,  3.87158096e-02,  5.02756275e-02, -8.95725042e-02,\n",
405 |               "        -1.45936176e-01, -9.95598454e-03, -4.72079590e-02,  3.96476686e-02,\n",
406 |               "        -2.22098202e-01, -7.88760409e-02, -2.44239885e-02,  8.99515226e-02],\n",
407 |               "       dtype=float32)]"
408 |             ]
409 |           },
410 |           "metadata": {
411 |             "tags": []
412 |           },
413 |           "execution_count": 40
414 |         }
415 |       ]
416 |     },
417 |     {
418 |       "metadata": {
419 |         "id": "4-RqzS88dgk_",
420 |         "colab_type": "text"
421 |       },
422 |       "cell_type": "markdown",
423 |       "source": [
424 |         "Salvando e carregando o modelo treinado"
425 |       ]
426 |     },
427 |     {
428 |       "metadata": {
429 |         "id": "SVjmU41IELzX",
430 |         "colab_type": "code",
431 |         "colab": {}
432 |       },
433 |       "cell_type": "code",
434 |       "source": [
435 |         "modelo.save('modelo.h5')\n",
436 |         "modelo_salvo = load_model('modelo.h5')"
437 |       ],
438 |       "execution_count": 0,
439 |       "outputs": []
440 |     },
441 |     {
442 |       "metadata": {
443 |         "id": "SVxHO3opdnMb",
444 |         "colab_type": "text"
445 |       },
446 |       "cell_type": "markdown",
447 |       "source": [
448 |         "Visualizando as acurácias de treino e validação por época"
449 |       ]
450 |     },
451 |     {
452 |       "metadata": {
453 |         "id": "pNc0JsWZY1Ie",
454 |         "colab_type": "code",
455 |         "colab": {
456 |           "base_uri": "https://localhost:8080/",
457 |           "height": 394
458 |         },
459 |         "outputId": "19182697-90e1-4179-928a-18b420c48d8a"
460 |       },
461 |       "cell_type": "code",
462 |       "source": [
463 |         "plt.plot(historico.history['acc'])\n",
464 |         "plt.plot(historico.history['val_acc'])\n",
465 |         "plt.title('Acurácia por épocas')\n",
466 |         "plt.xlabel('épocas')\n",
467 |         "plt.ylabel('acurácia')\n",
468 |         "plt.legend(['treino', 'validação'])\n"
469 |       ],
470 |       "execution_count": 8,
471 |       "outputs": [
472 |         {
473 |           "output_type": "execute_result",
474 |           "data": {
475 |             "text/plain": [
476 |               "<matplotlib.legend.Legend at 0x7f99cbdf73c8>"
477 |             ]
478 |           },
479 |           "metadata": {
480 |             "tags": []
481 |           },
482 |           "execution_count": 8
483 |         },
484 |         {
485 |           "output_type": "display_data",
486 |           "data": {
487 |             "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nERERi1Cpi4iIWMT/AT+ENTwu2bzXAAAAAElFTkSuQmCC\n",
488 |             "text/plain": [
489 |               "<matplotlib.figure.Figure at 0x7f99d28a4128>"
490 |             ]
491 |           },
492 |           "metadata": {
493 |             "tags": []
494 |           }
495 |         }
496 |       ]
497 |     },
498 |     {
499 |       "metadata": {
500 |         "id": "ZKxkWNawdzZh",
501 |         "colab_type": "text"
502 |       },
503 |       "cell_type": "markdown",
504 |       "source": [
505 |         "Visualizando as perdas de treino e validação por época"
506 |       ]
507 |     },
508 |     {
509 |       "metadata": {
510 |         "id": "2ugG3Vusg_Va",
511 |         "colab_type": "code",
512 |         "colab": {
513 |           "base_uri": "https://localhost:8080/",
514 |           "height": 393
515 |         },
516 |         "outputId": "dbc526a0-5890-4ece-a0e0-0ab01e18034a"
517 |       },
518 |       "cell_type": "code",
519 |       "source": [
520 |         "plt.plot(historico.history['loss'])\n",
521 |         "plt.plot(historico.history['val_loss'])\n",
522 |         "plt.title('Perda por épocas')\n",
523 |         "plt.xlabel('épocas')\n",
524 |         "plt.ylabel('perda')\n",
525 |         "plt.legend(['treino', 'validação'])"
526 |       ],
527 |       "execution_count": 9,
528 |       "outputs": [
529 |         {
530 |           "output_type": "execute_result",
531 |           "data": {
532 |             "text/plain": [
533 |               "<matplotlib.legend.Legend at 0x7f99cbd7bc18>"
534 |             ]
535 |           },
536 |           "metadata": {
537 |             "tags": []
538 |           },
539 |           "execution_count": 9
540 |         },
541 |         {
542 |           "output_type": "display_data",
543 |           "data": {
544 |             "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfsAAAFnCAYAAAChL+DqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
545 |             "text/plain": [
546 |               "<matplotlib.figure.Figure at 0x7f99cbdda630>"
547 |             ]
548 |           },
549 |           "metadata": {
550 |             "tags": []
551 |           }
552 |         }
553 |       ]
554 |     },
555 |     {
556 |       "metadata": {
557 |         "id": "JFXG1oibd8E8",
558 |         "colab_type": "text"
559 |       },
560 |       "cell_type": "markdown",
561 |       "source": [
562 |         "Testando o modelo e o modelo salvo"
563 |       ]
564 |     },
565 |     {
566 |       "metadata": {
567 |         "id": "I2kLEHPYUdhd",
568 |         "colab_type": "code",
569 |         "colab": {
570 |           "base_uri": "https://localhost:8080/",
571 |           "height": 85
572 |         },
573 |         "outputId": "c62dd99d-5620-437b-8cda-2553728c86c3"
574 |       },
575 |       "cell_type": "code",
576 |       "source": [
577 |         "testes = modelo.predict(imagens_teste)\n",
578 |         "print('resultado teste:', np.argmax(testes[1]))\n",
579 |         "print('número da imagem de teste:', identificacoes_teste[1])\n",
580 |         "\n",
581 |         "testes_modelo_salvo = modelo_salvo.predict(imagens_teste)\n",
582 |         "print('resultado teste modelo salvo:', np.argmax(testes_modelo_salvo[1]))\n",
583 |         "print('número da imagem de teste:', identificacoes_teste[1])"
584 |       ],
585 |       "execution_count": 10,
586 |       "outputs": [
587 |         {
588 |           "output_type": "stream",
589 |           "text": [
590 |             "resultado teste: 2\n",
591 |             "número da imagem de teste: 2\n",
592 |             "resultado teste modelo salvo: 2\n",
593 |             "número da imagem de teste: 2\n"
594 |           ],
595 |           "name": "stdout"
596 |         }
597 |       ]
598 |     },
599 |     {
600 |       "metadata": {
601 |         "id": "U64QMV1GeCUr",
602 |         "colab_type": "text"
603 |       },
604 |       "cell_type": "markdown",
605 |       "source": [
606 |         "Avaliando o modelo"
607 |       ]
608 |     },
609 |     {
610 |       "metadata": {
611 |         "id": "wsbr9WqyXQ2G",
612 |         "colab_type": "code",
613 |         "colab": {
614 |           "base_uri": "https://localhost:8080/",
615 |           "height": 68
616 |         },
617 |         "outputId": "cbf2e2c1-699b-4688-e770-e7ecbe5c0fd5"
618 |       },
619 |       "cell_type": "code",
620 |       "source": [
621 |         "perda_teste, acuracia_teste = modelo.evaluate(imagens_teste, identificacoes_teste)\n",
622 |         "print('Perda do teste:', perda_teste)\n",
623 |         "print('Acurácia do teste:', acuracia_teste)"
624 |       ],
625 |       "execution_count": 11,
626 |       "outputs": [
627 |         {
628 |           "output_type": "stream",
629 |           "text": [
630 |             "10000/10000 [==============================] - 0s 40us/step\n",
631 |             "Perda do teste: 2.2868376407623274\n",
632 |             "Acurácia do teste: 0.8565\n"
633 |           ],
634 |           "name": "stdout"
635 |         }
636 |       ]
637 |     }
638 |   ]
639 | }


--------------------------------------------------------------------------------
/Segundo_projeto_aula5.ipynb:
--------------------------------------------------------------------------------
1 | {"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"Segundo_projeto_aula5.ipynb","version":"0.3.2","provenance":[{"file_id":"1w-SdjiC3XEAlFe6PaYiFU-NL-c9sGroF","timestamp":1545061684493},{"file_id":"1XJcYr2kO2iMOxUFklsI8Nov8DHfWaN6F","timestamp":1545061633428},{"file_id":"1t03BioBUhAKHpvQmooWdnjxgkj-MAF7D","timestamp":1544203981880},{"file_id":"1-pv9T-KI09_jiMUe4MErOnAHR3IFfjtS","timestamp":1544114456244},{"file_id":"1Hg5iQPzIw9n7ioWeD9Nv3AnAPFDA0bE_","timestamp":1538632904381}],"collapsed_sections":[]},"kernelspec":{"name":"python3","display_name":"Python 3"}},"cells":[{"metadata":{"id":"sNc3NouWYrN7","colab_type":"text"},"cell_type":"markdown","source":["Imports"]},{"metadata":{"id":"4bQCahYjYdUB","colab_type":"code","colab":{}},"cell_type":"code","source":["import tensorflow\n","from tensorflow import keras\n","import matplotlib.pyplot as plt\n","import numpy as np\n","from tensorflow.keras.models import load_model"],"execution_count":0,"outputs":[]},{"metadata":{"id":"f8uuzUjVYvkf","colab_type":"text"},"cell_type":"markdown","source":["Carregando o dataset"]},{"metadata":{"id":"QoIys3wMYirH","colab_type":"code","colab":{}},"cell_type":"code","source":["dataset = keras.datasets.fashion_mnist\n","((imagens_treino, identificacoes_treino), (imagens_teste, identificacoes_teste)) = dataset.load_data()\n"],"execution_count":0,"outputs":[]},{"metadata":{"id":"oFr0526ZY5E_","colab_type":"text"},"cell_type":"markdown","source":["Exploração dos dados"]},{"metadata":{"id":"fPIT36hpYl5A","colab_type":"code","outputId":"812c2fcc-0c28-431b-91b8-1e64a05f31e8","executionInfo":{"status":"ok","timestamp":1545137955774,"user_tz":120,"elapsed":1479,"user":{"displayName":"Cássia Sampaio","photoUrl":"https://lh5.googleusercontent.com/-WZZI20Ykt7M/AAAAAAAAAAI/AAAAAAAAAlI/NPtGozqQ-38/s64/photo.jpg","userId":"11135730660253677021"}},"colab":{"base_uri":"https://localhost:8080/","height":35}},"cell_type":"code","source":["len(imagens_treino)\n","imagens_treino.shape\n","imagens_teste.shape\n","len(identificacoes_teste)\n","identificacoes_treino.min()\n","identificacoes_treino.max()"],"execution_count":11,"outputs":[{"output_type":"execute_result","data":{"text/plain":["9"]},"metadata":{"tags":[]},"execution_count":11}]},{"metadata":{"id":"yTWqT9DIY-iB","colab_type":"text"},"cell_type":"markdown","source":["Exibição dos dados"]},{"metadata":{"id":"bGESm49JVahh","colab_type":"code","outputId":"25be141d-aa74-45fc-da00-f38b9a8f3039","executionInfo":{"status":"ok","timestamp":1545137956003,"user_tz":120,"elapsed":1699,"user":{"displayName":"Cássia Sampaio","photoUrl":"https://lh5.googleusercontent.com/-WZZI20Ykt7M/AAAAAAAAAAI/AAAAAAAAAlI/NPtGozqQ-38/s64/photo.jpg","userId":"11135730660253677021"}},"colab":{"base_uri":"https://localhost:8080/","height":365}},"cell_type":"code","source":["total_de_classificacoes = 10\n","nomes_de_classificacoes = ['Camiseta', 'Calça', 'Pullover', \n","                           'Vestido', 'Casaco', 'Sandália', 'Camisa',\n","                           'Tênis', 'Bolsa', 'Bota']\n","\n","plt.imshow(imagens_treino[0])\n","plt.colorbar()"],"execution_count":12,"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.colorbar.Colorbar at 0x7f3642983ef0>"]},"metadata":{"tags":[]},"execution_count":12},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAYgAAAFKCAYAAAAHY8iZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n","text/plain":["<matplotlib.figure.Figure at 0x7f364525a9e8>"]},"metadata":{"tags":[]}}]},{"metadata":{"id":"uU6ZySpIcoc5","colab_type":"text"},"cell_type":"markdown","source":["Normalizando as imagens"]},{"metadata":{"id":"Ral_hdl9ulGG","colab_type":"code","colab":{}},"cell_type":"code","source":["imagens_treino = imagens_treino/float(255)"],"execution_count":0,"outputs":[]},{"metadata":{"id":"2WABnvOJdCNl","colab_type":"text"},"cell_type":"markdown","source":["Criando, compilando, treinando e normalizando o modelo"]},{"metadata":{"id":"uFedwlfFc0ii","colab_type":"code","outputId":"b2115202-893b-4830-d587-5b6608639925","executionInfo":{"status":"ok","timestamp":1545143728581,"user_tz":120,"elapsed":11713,"user":{"displayName":"Cássia Sampaio","photoUrl":"https://lh5.googleusercontent.com/-WZZI20Ykt7M/AAAAAAAAAAI/AAAAAAAAAlI/NPtGozqQ-38/s64/photo.jpg","userId":"11135730660253677021"}},"colab":{"base_uri":"https://localhost:8080/","height":228}},"cell_type":"code","source":["modelo = keras.Sequential([ \n","    keras.layers.Flatten(input_shape=(28, 28)),\n","    keras.layers.Dense(256, activation=tensorflow.nn.relu),\n","    keras.layers.Dropout(0.2),\n","    keras.layers.Dense(10, activation=tensorflow.nn.softmax)\n","])\n","\n","adam = keras.optimizers.Adam(lr=0.002)\n","\n","'''lr\n","learning rate\n","taxa de aprendizado'''\n","\n","parando_cedo = [keras.callbacks.EarlyStopping(monitor='val_loss'),\n","                keras.callbacks.ModelCheckpoint(filepath='melhor_modelo.hdf5',\n","                                               monitor='val_loss', save_best_only=True)]\n","\n","modelo.compile(optimizer=adam, \n","               loss='sparse_categorical_crossentropy',\n","               metrics=['accuracy'])\n","\n","historico = modelo.fit(imagens_treino, identificacoes_treino, \n","                       batch_size = 480,\n","                       epochs=5, validation_split=0.2,\n","                       callbacks=parando_cedo)"],"execution_count":24,"outputs":[{"output_type":"stream","text":["Train on 48000 samples, validate on 12000 samples\n","Epoch 1/5\n","48000/48000 [==============================] - 3s 60us/step - loss: 0.6426 - acc: 0.7757 - val_loss: 0.4523 - val_acc: 0.8387\n","Epoch 2/5\n","48000/48000 [==============================] - 2s 40us/step - loss: 0.4276 - acc: 0.8485 - val_loss: 0.3967 - val_acc: 0.8597\n","Epoch 3/5\n","48000/48000 [==============================] - 2s 40us/step - loss: 0.3864 - acc: 0.8630 - val_loss: 0.3775 - val_acc: 0.8618\n","Epoch 4/5\n","48000/48000 [==============================] - 2s 40us/step - loss: 0.3603 - acc: 0.8719 - val_loss: 0.3512 - val_acc: 0.8747\n","Epoch 5/5\n","48000/48000 [==============================] - 2s 41us/step - loss: 0.3418 - acc: 0.8760 - val_loss: 0.3430 - val_acc: 0.8802\n"],"name":"stdout"}]},{"metadata":{"id":"sd2k5eXndvza","colab_type":"text"},"cell_type":"markdown","source":[""]},{"metadata":{"id":"weeyzt2QuDXS","colab_type":"text"},"cell_type":"markdown","source":["Sumário do modelo"]},{"metadata":{"colab_type":"code","outputId":"cb5ca991-99c3-41e8-d71a-81dedae12cb4","executionInfo":{"status":"ok","timestamp":1545138006111,"user_tz":120,"elapsed":51783,"user":{"displayName":"Cássia Sampaio","photoUrl":"https://lh5.googleusercontent.com/-WZZI20Ykt7M/AAAAAAAAAAI/AAAAAAAAAlI/NPtGozqQ-38/s64/photo.jpg","userId":"11135730660253677021"}},"id":"QLG_CH6Lr2vp","colab":{"base_uri":"https://localhost:8080/","height":277}},"cell_type":"code","source":["sumario_do_modelo = modelo.summary()"],"execution_count":15,"outputs":[{"output_type":"stream","text":["_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","flatten (Flatten)            (None, 784)               0         \n","_________________________________________________________________\n","dense (Dense)                (None, 256)               200960    \n","_________________________________________________________________\n","dropout (Dropout)            (None, 256)               0         \n","_________________________________________________________________\n","dense_1 (Dense)              (None, 10)                2570      \n","=================================================================\n","Total params: 203,530\n","Trainable params: 203,530\n","Non-trainable params: 0\n","_________________________________________________________________\n"],"name":"stdout"}]},{"metadata":{"id":"GVuSQg42udch","colab_type":"text"},"cell_type":"markdown","source":["Pesos da primeira camada Dense"]},{"metadata":{"id":"YRzqJi5gr-ug","colab_type":"code","outputId":"acb8a25a-268c-4a93-da4c-697b0e6c071b","executionInfo":{"status":"ok","timestamp":1545138006114,"user_tz":120,"elapsed":51780,"user":{"displayName":"Cássia Sampaio","photoUrl":"https://lh5.googleusercontent.com/-WZZI20Ykt7M/AAAAAAAAAAI/AAAAAAAAAlI/NPtGozqQ-38/s64/photo.jpg","userId":"11135730660253677021"}},"colab":{"base_uri":"https://localhost:8080/","height":243}},"cell_type":"code","source":["pesos_camada_dense = modelo.layers[1].get_weights()[0]\n","pesos_camada_dense"],"execution_count":16,"outputs":[{"output_type":"execute_result","data":{"text/plain":["array([[ 6.1770529e-04, -5.7177022e-03,  4.9233064e-03, ...,\n","        -3.4757730e-02, -2.9884566e-02, -5.8418039e-02],\n","       [-7.0627712e-02, -3.2129511e-03,  6.4527035e-02, ...,\n","         7.3822394e-02,  7.4995592e-02, -5.1243432e-02],\n","       [ 1.2629986e-02, -7.5908080e-02, -2.6841402e+00, ...,\n","        -4.8490271e-02,  6.2874153e-02,  2.5197577e+00],\n","       ...,\n","       [ 1.0123365e-02, -3.8167909e-02, -2.8177865e+00, ...,\n","        -2.8139663e+00, -2.6680083e+00,  2.6510463e+00],\n","       [ 1.5732393e-02, -2.7290877e-02, -1.0481872e-02, ...,\n","        -1.5448462e-02,  1.1646114e-02,  2.9921636e-02],\n","       [-2.4638295e-02,  2.6490360e-02,  6.6962734e-02, ...,\n","         2.7110979e-02, -4.3123923e-02,  2.5531299e-02]], dtype=float32)"]},"metadata":{"tags":[]},"execution_count":16}]},{"metadata":{"id":"k5vzzzzF-pdh","colab_type":"code","outputId":"8e285466-cda4-4e6b-c4a2-b2ba77f882d7","executionInfo":{"status":"error","timestamp":1545138007320,"user_tz":120,"elapsed":52970,"user":{"displayName":"Cássia Sampaio","photoUrl":"https://lh5.googleusercontent.com/-WZZI20Ykt7M/AAAAAAAAAAI/AAAAAAAAAlI/NPtGozqQ-38/s64/photo.jpg","userId":"11135730660253677021"}},"colab":{"base_uri":"https://localhost:8080/","height":235}},"cell_type":"code","source":["pesos_camada_dense.shape\n","pesos_camada_dense_zerados = np.zeros((784, 256))\n","\n","pesos_camada_dense_aleatorios = np.random.rand(784, 256)\n","\n","'''\n","modelo.layers[1].set_weights(\n","    [pesos_camada_dense_zerados, vieses_camada_dense]\n",")\n","'''\n","\n","modelo.layers[1].set_weights(\n","    [pesos_camada_dense_aleatorios, vieses_camada_dense]\n",")\n","\n","modelo.layers[1].get_weights()\n"],"execution_count":17,"outputs":[{"output_type":"error","ename":"NameError","evalue":"ignored","traceback":["\u001b[0;31m---------------------------------------------------------------------------\u001b[0m","\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)","\u001b[0;32m<ipython-input-17-acd4227923cf>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     11\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m modelo.layers[1].set_weights(\n\u001b[0;32m---> 13\u001b[0;31m     \u001b[0;34m[\u001b[0m\u001b[0mpesos_camada_dense_aleatorios\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvieses_camada_dense\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     14\u001b[0m )\n\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n","\u001b[0;31mNameError\u001b[0m: name 'vieses_camada_dense' is not defined"]}]},{"metadata":{"id":"0nJSiAyI_-3a","colab_type":"code","colab":{}},"cell_type":"code","source":["modelo.compile(optimizer='adam', \n","               loss='sparse_categorical_crossentropy',\n","               metrics=['accuracy'])\n","\n","historico = modelo.fit(imagens_treino, identificacoes_treino, epochs=5, validation_split=0.2)"],"execution_count":0,"outputs":[]},{"metadata":{"id":"TqiDVI-aAi9k","colab_type":"code","colab":{}},"cell_type":"code","source":["plt.plot(historico.history['acc'])\n","plt.plot(historico.history['val_acc'])\n","plt.title('Acurácia por épocas')\n","plt.xlabel('épocas')\n","plt.ylabel('acurácia')\n","plt.legend(['treino', 'validação'])\n"],"execution_count":0,"outputs":[]},{"metadata":{"id":"XIYEY-hNEYT7","colab_type":"code","colab":{}},"cell_type":"code","source":["plt.plot(historico.history['acc'])\n","plt.plot(historico.history['val_acc'])\n","plt.title('Acurácia por épocas')\n","plt.xlabel('épocas')\n","plt.ylabel('acurácia')\n","plt.legend(['treino', 'validação'])"],"execution_count":0,"outputs":[]},{"metadata":{"id":"sELsC7XkFCei","colab_type":"code","outputId":"1de3e550-e52b-4e0c-fff5-cb1981361c3c","executionInfo":{"status":"ok","timestamp":1545144456753,"user_tz":120,"elapsed":597,"user":{"displayName":"Cássia Sampaio","photoUrl":"https://lh5.googleusercontent.com/-WZZI20Ykt7M/AAAAAAAAAAI/AAAAAAAAAlI/NPtGozqQ-38/s64/photo.jpg","userId":"11135730660253677021"}},"colab":{"base_uri":"https://localhost:8080/","height":815}},"cell_type":"code","source":["modelo.get_config()"],"execution_count":25,"outputs":[{"output_type":"execute_result","data":{"text/plain":["{'layers': [{'class_name': 'Flatten',\n","   'config': {'batch_input_shape': (None, 28, 28),\n","    'data_format': 'channels_last',\n","    'dtype': 'float32',\n","    'name': 'flatten_6',\n","    'trainable': True}},\n","  {'class_name': 'Dense',\n","   'config': {'activation': 'relu',\n","    'activity_regularizer': None,\n","    'bias_constraint': None,\n","    'bias_initializer': {'class_name': 'Zeros',\n","     'config': {'dtype': 'float32'}},\n","    'bias_regularizer': None,\n","    'dtype': 'float32',\n","    'kernel_constraint': None,\n","    'kernel_initializer': {'class_name': 'GlorotUniform',\n","     'config': {'dtype': 'float32', 'seed': None}},\n","    'kernel_regularizer': None,\n","    'name': 'dense_12',\n","    'trainable': True,\n","    'units': 256,\n","    'use_bias': True}},\n","  {'class_name': 'Dropout',\n","   'config': {'dtype': 'float32',\n","    'name': 'dropout_6',\n","    'noise_shape': None,\n","    'rate': 0.2,\n","    'seed': None,\n","    'trainable': True}},\n","  {'class_name': 'Dense',\n","   'config': {'activation': 'softmax',\n","    'activity_regularizer': None,\n","    'bias_constraint': None,\n","    'bias_initializer': {'class_name': 'Zeros',\n","     'config': {'dtype': 'float32'}},\n","    'bias_regularizer': None,\n","    'dtype': 'float32',\n","    'kernel_constraint': None,\n","    'kernel_initializer': {'class_name': 'GlorotUniform',\n","     'config': {'dtype': 'float32', 'seed': None}},\n","    'kernel_regularizer': None,\n","    'name': 'dense_13',\n","    'trainable': True,\n","    'units': 10,\n","    'use_bias': True}}],\n"," 'name': 'sequential_6'}"]},"metadata":{"tags":[]},"execution_count":25}]},{"metadata":{"id":"4S8FoWQ1Ar_x","colab_type":"code","colab":{}},"cell_type":"code","source":["plt.plot(historico.history['loss'])\n","plt.plot(historico.history['val_loss'])\n","plt.title('Perda por épocas')\n","plt.xlabel('épocas')\n","plt.ylabel('perda')\n","plt.legend(['treino', 'validação'])"],"execution_count":0,"outputs":[]},{"metadata":{"id":"Vuve5QcuF4eE","colab_type":"code","colab":{}},"cell_type":"code","source":["vieses_camada_dense = modelo.layers[1].get_weights()[1]\n","vieses_camada_dense.shape"],"execution_count":0,"outputs":[]},{"metadata":{"id":"5lEZYFfHHfrx","colab_type":"code","colab":{}},"cell_type":"code","source":["vieses_camada_dense_zerados = np.zeros((256,))\n","modelo.layers[1].set_weights(\n","    [pesos_camada_dense_aleatorios, vieses_camada_dense_zerados]\n",")"],"execution_count":0,"outputs":[]},{"metadata":{"id":"4-RqzS88dgk_","colab_type":"text"},"cell_type":"markdown","source":["Salvando e carregando o modelo treinado"]},{"metadata":{"id":"SVjmU41IELzX","colab_type":"code","colab":{}},"cell_type":"code","source":["modelo.save('modelo.h5')\n","modelo_salvo = load_model('modelo.h5')"],"execution_count":0,"outputs":[]},{"metadata":{"id":"SVxHO3opdnMb","colab_type":"text"},"cell_type":"markdown","source":["Visualizando as acurácias de treino e validação por época"]},{"metadata":{"id":"pNc0JsWZY1Ie","colab_type":"code","colab":{}},"cell_type":"code","source":["plt.plot(historico.history['acc'])\n","plt.plot(historico.history['val_acc'])\n","plt.title('Acurácia por épocas')\n","plt.xlabel('épocas')\n","plt.ylabel('acurácia')\n","plt.legend(['treino', 'validação'])\n"],"execution_count":0,"outputs":[]},{"metadata":{"id":"ZKxkWNawdzZh","colab_type":"text"},"cell_type":"markdown","source":["Visualizando as perdas de treino e validação por época"]},{"metadata":{"id":"2ugG3Vusg_Va","colab_type":"code","colab":{}},"cell_type":"code","source":["plt.plot(historico.history['loss'])\n","plt.plot(historico.history['val_loss'])\n","plt.title('Perda por épocas')\n","plt.xlabel('épocas')\n","plt.ylabel('perda')\n","plt.legend(['treino', 'validação'])"],"execution_count":0,"outputs":[]},{"metadata":{"id":"JFXG1oibd8E8","colab_type":"text"},"cell_type":"markdown","source":["Testando o modelo e o modelo salvo"]},{"metadata":{"id":"I2kLEHPYUdhd","colab_type":"code","colab":{}},"cell_type":"code","source":["testes = modelo.predict(imagens_teste)\n","print('resultado teste:', np.argmax(testes[1]))\n","print('número da imagem de teste:', identificacoes_teste[1])\n","\n","testes_modelo_salvo = modelo_salvo.predict(imagens_teste)\n","print('resultado teste modelo salvo:', np.argmax(testes_modelo_salvo[1]))\n","print('número da imagem de teste:', identificacoes_teste[1])"],"execution_count":0,"outputs":[]},{"metadata":{"id":"U64QMV1GeCUr","colab_type":"text"},"cell_type":"markdown","source":["Avaliando o modelo"]},{"metadata":{"id":"wsbr9WqyXQ2G","colab_type":"code","colab":{}},"cell_type":"code","source":["perda_teste, acuracia_teste = modelo.evaluate(imagens_teste, identificacoes_teste)\n","print('Perda do teste:', perda_teste)\n","print('Acurácia do teste:', acuracia_teste)"],"execution_count":0,"outputs":[]}]}


--------------------------------------------------------------------------------