Skip to content

Instantly share code, notes, and snippets.

@mynameisfiber

mynameisfiber/IRIS.ipynb

Created January 12, 2017 22:34
Show Gist options
  • Save mynameisfiber/da387afc03b8768b8aed993d59752bde to your computer and use it in GitHub Desktop.
Save mynameisfiber/da387afc03b8768b8aed993d59752bde to your computer and use it in GitHub Desktop.
Download ZIP
IRIS with keras
Display the source blob
Display the rendered blob
Raw
IRIS.ipynb
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Using Theano backend.\n"
]
}
],
"source": [
"from sklearn import datasets\n",
"\n",
"from keras.models import Sequential\n",
"from keras.layers import Dense\n",
"from keras.utils import np_utils\n",
"from keras.callbacks import EarlyStopping\n",
"\n",
"import numpy as np\n",
"np.random.seed(1337)\n",
"\n",
"import pylab as py\n",
"import seaborn as sns\n",
"%matplotlib inline"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"def plot_history(history):\n",
" interesting_measures = [k for k in history.history.keys()\n",
" if not k.startswith('val_')]\n",
" fig, axs = py.subplots(len(interesting_measures))\n",
" for axs, measure in zip(axs.flatten(), interesting_measures):\n",
" axs.set_ylabel(measure)\n",
" axs.plot(history.epoch, history.history['val_' + measure], label='val_' + measure)\n",
" axs.plot(history.epoch, history.history[measure], label=measure)\n",
" axs.legend()\n",
" return fig"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"collapsed": false
},
"outputs": [],
"source": [
"iris = datasets.load_iris()\n",
"\n",
"X = iris.data\n",
"y = np_utils.to_categorical(iris.target).astype('uint8')\n",
"\n",
"shuffle_idx = np.random.permutation(X.shape[0])\n",
"X = X[shuffle_idx]\n",
"y = y[shuffle_idx]\n",
"\n",
"nb_features = X.shape[1]\n",
"nb_classes = y.shape[1]"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"____________________________________________________________________________________________________\n",
"Layer (type) Output Shape Param # Connected to \n",
"====================================================================================================\n",
"dense_1 (Dense) (None, 3) 15 dense_input_1[0][0] \n",
"====================================================================================================\n",
"Total params: 15\n",
"Trainable params: 15\n",
"Non-trainable params: 0\n",
"____________________________________________________________________________________________________\n"
]
}
],
"source": [
"model = Sequential()\n",
"model.add(Dense(nb_classes, input_dim=nb_features, activation='softmax'))\n",
"\n",
"model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])\n",
"model.summary()"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Train on 75 samples, validate on 75 samples\n",
"Epoch 1/10000\n",
"0s - loss: 0.9731 - acc: 0.3333 - val_loss: 0.8665 - val_acc: 0.4400\n",
"Epoch 2/10000\n",
"0s - loss: 0.9028 - acc: 0.4267 - val_loss: 0.8258 - val_acc: 0.4533\n",
"Epoch 3/10000\n",
"0s - loss: 0.8464 - acc: 0.5600 - val_loss: 0.7936 - val_acc: 0.6133\n",
"Epoch 4/10000\n",
"0s - loss: 0.7971 - acc: 0.7067 - val_loss: 0.7718 - val_acc: 0.6933\n",
"Epoch 5/10000\n",
"0s - loss: 0.7625 - acc: 0.7467 - val_loss: 0.7577 - val_acc: 0.7467\n",
"Epoch 6/10000\n",
"0s - loss: 0.7370 - acc: 0.7733 - val_loss: 0.7457 - val_acc: 0.7200\n",
"Epoch 7/10000\n",
"0s - loss: 0.7188 - acc: 0.7600 - val_loss: 0.7366 - val_acc: 0.6933\n",
"Epoch 8/10000\n",
"0s - loss: 0.7032 - acc: 0.7867 - val_loss: 0.7298 - val_acc: 0.7200\n",
"Epoch 9/10000\n",
"0s - loss: 0.6928 - acc: 0.7733 - val_loss: 0.7233 - val_acc: 0.7600\n",
"Epoch 10/10000\n",
"0s - loss: 0.6792 - acc: 0.7867 - val_loss: 0.7152 - val_acc: 0.7333\n",
"Epoch 11/10000\n",
"0s - loss: 0.6697 - acc: 0.7600 - val_loss: 0.7087 - val_acc: 0.6800\n",
"Epoch 12/10000\n",
"0s - loss: 0.6608 - acc: 0.7200 - val_loss: 0.7019 - val_acc: 0.6933\n",
"Epoch 13/10000\n",
"0s - loss: 0.6559 - acc: 0.8000 - val_loss: 0.6963 - val_acc: 0.7467\n",
"Epoch 14/10000\n",
"0s - loss: 0.6454 - acc: 0.7467 - val_loss: 0.6878 - val_acc: 0.6800\n",
"Epoch 15/10000\n",
"0s - loss: 0.6353 - acc: 0.7200 - val_loss: 0.6825 - val_acc: 0.6933\n",
"Epoch 16/10000\n",
"0s - loss: 0.6283 - acc: 0.7600 - val_loss: 0.6761 - val_acc: 0.7600\n",
"Epoch 17/10000\n",
"0s - loss: 0.6215 - acc: 0.7867 - val_loss: 0.6716 - val_acc: 0.7467\n",
"Epoch 18/10000\n",
"0s - loss: 0.6145 - acc: 0.7733 - val_loss: 0.6654 - val_acc: 0.7333\n",
"Epoch 19/10000\n",
"0s - loss: 0.6077 - acc: 0.8133 - val_loss: 0.6598 - val_acc: 0.7467\n",
"Epoch 20/10000\n",
"0s - loss: 0.6033 - acc: 0.7467 - val_loss: 0.6541 - val_acc: 0.7333\n",
"Epoch 21/10000\n",
"0s - loss: 0.5947 - acc: 0.8267 - val_loss: 0.6496 - val_acc: 0.7600\n",
"Epoch 22/10000\n",
"0s - loss: 0.5906 - acc: 0.8133 - val_loss: 0.6455 - val_acc: 0.7600\n",
"Epoch 23/10000\n",
"0s - loss: 0.5842 - acc: 0.8000 - val_loss: 0.6397 - val_acc: 0.7867\n",
"Epoch 24/10000\n",
"0s - loss: 0.5787 - acc: 0.8133 - val_loss: 0.6347 - val_acc: 0.7733\n",
"Epoch 25/10000\n",
"0s - loss: 0.5735 - acc: 0.8267 - val_loss: 0.6306 - val_acc: 0.7867\n",
"Epoch 26/10000\n",
"0s - loss: 0.5684 - acc: 0.8133 - val_loss: 0.6260 - val_acc: 0.7867\n",
"Epoch 27/10000\n",
"0s - loss: 0.5673 - acc: 0.7733 - val_loss: 0.6217 - val_acc: 0.7200\n",
"Epoch 28/10000\n",
"0s - loss: 0.5577 - acc: 0.8533 - val_loss: 0.6177 - val_acc: 0.8000\n",
"Epoch 29/10000\n",
"0s - loss: 0.5550 - acc: 0.8133 - val_loss: 0.6139 - val_acc: 0.8000\n",
"Epoch 30/10000\n",
"0s - loss: 0.5503 - acc: 0.8400 - val_loss: 0.6109 - val_acc: 0.8000\n",
"Epoch 31/10000\n",
"0s - loss: 0.5453 - acc: 0.8133 - val_loss: 0.6061 - val_acc: 0.8133\n",
"Epoch 32/10000\n",
"0s - loss: 0.5411 - acc: 0.8533 - val_loss: 0.6021 - val_acc: 0.8133\n",
"Epoch 33/10000\n",
"0s - loss: 0.5370 - acc: 0.8533 - val_loss: 0.5984 - val_acc: 0.8133\n",
"Epoch 34/10000\n",
"0s - loss: 0.5336 - acc: 0.8400 - val_loss: 0.5949 - val_acc: 0.7867\n",
"Epoch 35/10000\n",
"0s - loss: 0.5299 - acc: 0.8267 - val_loss: 0.5913 - val_acc: 0.8133\n",
"Epoch 36/10000\n",
"0s - loss: 0.5260 - acc: 0.8400 - val_loss: 0.5878 - val_acc: 0.8133\n",
"Epoch 37/10000\n",
"0s - loss: 0.5218 - acc: 0.8533 - val_loss: 0.5849 - val_acc: 0.8000\n",
"Epoch 38/10000\n",
"0s - loss: 0.5195 - acc: 0.8267 - val_loss: 0.5814 - val_acc: 0.8000\n",
"Epoch 39/10000\n",
"0s - loss: 0.5169 - acc: 0.8400 - val_loss: 0.5790 - val_acc: 0.8267\n",
"Epoch 40/10000\n",
"0s - loss: 0.5113 - acc: 0.8533 - val_loss: 0.5760 - val_acc: 0.8133\n",
"Epoch 41/10000\n",
"0s - loss: 0.5098 - acc: 0.8533 - val_loss: 0.5721 - val_acc: 0.8267\n",
"Epoch 42/10000\n",
"0s - loss: 0.5060 - acc: 0.8667 - val_loss: 0.5693 - val_acc: 0.8267\n",
"Epoch 43/10000\n",
"0s - loss: 0.5059 - acc: 0.8267 - val_loss: 0.5675 - val_acc: 0.7867\n",
"Epoch 44/10000\n",
"0s - loss: 0.4991 - acc: 0.8400 - val_loss: 0.5638 - val_acc: 0.8133\n",
"Epoch 45/10000\n",
"0s - loss: 0.4964 - acc: 0.8400 - val_loss: 0.5607 - val_acc: 0.8133\n",
"Epoch 46/10000\n",
"0s - loss: 0.4933 - acc: 0.8667 - val_loss: 0.5586 - val_acc: 0.8267\n",
"Epoch 47/10000\n",
"0s - loss: 0.4918 - acc: 0.8667 - val_loss: 0.5558 - val_acc: 0.8267\n",
"Epoch 48/10000\n",
"0s - loss: 0.4892 - acc: 0.8667 - val_loss: 0.5533 - val_acc: 0.8267\n",
"Epoch 49/10000\n",
"0s - loss: 0.4856 - acc: 0.8667 - val_loss: 0.5505 - val_acc: 0.8400\n",
"Epoch 50/10000\n",
"0s - loss: 0.4831 - acc: 0.8800 - val_loss: 0.5486 - val_acc: 0.8267\n",
"Epoch 51/10000\n",
"0s - loss: 0.4810 - acc: 0.8933 - val_loss: 0.5459 - val_acc: 0.8267\n",
"Epoch 52/10000\n",
"0s - loss: 0.4775 - acc: 0.8667 - val_loss: 0.5435 - val_acc: 0.8400\n",
"Epoch 53/10000\n",
"0s - loss: 0.4753 - acc: 0.8667 - val_loss: 0.5413 - val_acc: 0.8400\n",
"Epoch 54/10000\n",
"0s - loss: 0.4731 - acc: 0.8800 - val_loss: 0.5388 - val_acc: 0.8267\n",
"Epoch 55/10000\n",
"0s - loss: 0.4698 - acc: 0.8667 - val_loss: 0.5364 - val_acc: 0.8267\n",
"Epoch 56/10000\n",
"0s - loss: 0.4681 - acc: 0.8800 - val_loss: 0.5347 - val_acc: 0.8400\n",
"Epoch 57/10000\n",
"0s - loss: 0.4653 - acc: 0.8800 - val_loss: 0.5320 - val_acc: 0.8400\n",
"Epoch 58/10000\n",
"0s - loss: 0.4653 - acc: 0.8667 - val_loss: 0.5306 - val_acc: 0.8267\n",
"Epoch 59/10000\n",
"0s - loss: 0.4608 - acc: 0.8800 - val_loss: 0.5281 - val_acc: 0.8400\n",
"Epoch 60/10000\n",
"0s - loss: 0.4596 - acc: 0.8800 - val_loss: 0.5266 - val_acc: 0.8667\n",
"Epoch 61/10000\n",
"0s - loss: 0.4572 - acc: 0.8933 - val_loss: 0.5245 - val_acc: 0.8267\n",
"Epoch 62/10000\n",
"0s - loss: 0.4544 - acc: 0.8933 - val_loss: 0.5217 - val_acc: 0.8400\n",
"Epoch 63/10000\n",
"0s - loss: 0.4529 - acc: 0.8800 - val_loss: 0.5200 - val_acc: 0.8400\n",
"Epoch 64/10000\n",
"0s - loss: 0.4505 - acc: 0.8800 - val_loss: 0.5179 - val_acc: 0.8400\n",
"Epoch 65/10000\n",
"0s - loss: 0.4489 - acc: 0.8933 - val_loss: 0.5160 - val_acc: 0.8400\n",
"Epoch 66/10000\n",
"0s - loss: 0.4468 - acc: 0.8800 - val_loss: 0.5141 - val_acc: 0.8400\n",
"Epoch 67/10000\n",
"0s - loss: 0.4464 - acc: 0.8933 - val_loss: 0.5128 - val_acc: 0.8533\n",
"Epoch 68/10000\n",
"0s - loss: 0.4431 - acc: 0.8933 - val_loss: 0.5105 - val_acc: 0.8533\n",
"Epoch 69/10000\n",
"0s - loss: 0.4407 - acc: 0.8800 - val_loss: 0.5087 - val_acc: 0.8400\n",
"Epoch 70/10000\n",
"0s - loss: 0.4392 - acc: 0.8800 - val_loss: 0.5069 - val_acc: 0.8533\n",
"Epoch 71/10000\n",
"0s - loss: 0.4381 - acc: 0.8800 - val_loss: 0.5054 - val_acc: 0.8400\n",
"Epoch 72/10000\n",
"0s - loss: 0.4363 - acc: 0.8933 - val_loss: 0.5036 - val_acc: 0.8400\n",
"Epoch 73/10000\n",
"0s - loss: 0.4347 - acc: 0.8933 - val_loss: 0.5016 - val_acc: 0.8400\n",
"Epoch 74/10000\n",
"0s - loss: 0.4320 - acc: 0.8933 - val_loss: 0.4999 - val_acc: 0.8400\n",
"Epoch 75/10000\n",
"0s - loss: 0.4302 - acc: 0.9067 - val_loss: 0.4987 - val_acc: 0.8400\n",
"Epoch 76/10000\n",
"0s - loss: 0.4289 - acc: 0.8933 - val_loss: 0.4967 - val_acc: 0.8400\n",
"Epoch 77/10000\n",
"0s - loss: 0.4271 - acc: 0.8933 - val_loss: 0.4950 - val_acc: 0.8533\n",
"Epoch 78/10000\n",
"0s - loss: 0.4252 - acc: 0.9067 - val_loss: 0.4932 - val_acc: 0.8400\n",
"Epoch 79/10000\n",
"0s - loss: 0.4235 - acc: 0.9067 - val_loss: 0.4917 - val_acc: 0.8533\n",
"Epoch 80/10000\n",
"0s - loss: 0.4226 - acc: 0.9067 - val_loss: 0.4901 - val_acc: 0.8400\n",
"Epoch 81/10000\n",
"0s - loss: 0.4204 - acc: 0.9067 - val_loss: 0.4888 - val_acc: 0.8400\n",
"Epoch 82/10000\n",
"0s - loss: 0.4194 - acc: 0.8933 - val_loss: 0.4876 - val_acc: 0.8533\n",
"Epoch 83/10000\n",
"0s - loss: 0.4180 - acc: 0.8933 - val_loss: 0.4860 - val_acc: 0.8533\n",
"Epoch 84/10000\n",
"0s - loss: 0.4175 - acc: 0.9067 - val_loss: 0.4846 - val_acc: 0.8667\n",
"Epoch 85/10000\n",
"0s - loss: 0.4151 - acc: 0.9067 - val_loss: 0.4831 - val_acc: 0.8667\n",
"Epoch 86/10000\n",
"0s - loss: 0.4144 - acc: 0.9067 - val_loss: 0.4817 - val_acc: 0.8533\n",
"Epoch 87/10000\n",
"0s - loss: 0.4120 - acc: 0.8933 - val_loss: 0.4800 - val_acc: 0.8667\n",
"Epoch 88/10000\n",
"0s - loss: 0.4105 - acc: 0.9067 - val_loss: 0.4784 - val_acc: 0.8667\n",
"Epoch 89/10000\n",
"0s - loss: 0.4096 - acc: 0.9067 - val_loss: 0.4773 - val_acc: 0.8800\n",
"Epoch 90/10000\n",
"0s - loss: 0.4104 - acc: 0.9200 - val_loss: 0.4767 - val_acc: 0.9333\n",
"Epoch 91/10000\n",
"0s - loss: 0.4084 - acc: 0.9200 - val_loss: 0.4748 - val_acc: 0.8533\n",
"Epoch 92/10000\n",
"0s - loss: 0.4094 - acc: 0.9067 - val_loss: 0.4736 - val_acc: 0.8933\n",
"Epoch 93/10000\n",
"0s - loss: 0.4038 - acc: 0.9200 - val_loss: 0.4717 - val_acc: 0.8933\n",
"Epoch 94/10000\n",
"0s - loss: 0.4024 - acc: 0.9200 - val_loss: 0.4707 - val_acc: 0.8800\n",
"Epoch 95/10000\n",
"0s - loss: 0.4022 - acc: 0.9333 - val_loss: 0.4691 - val_acc: 0.8800\n",
"Epoch 96/10000\n",
"0s - loss: 0.3998 - acc: 0.9200 - val_loss: 0.4678 - val_acc: 0.8933\n",
"Epoch 97/10000\n",
"0s - loss: 0.3985 - acc: 0.9200 - val_loss: 0.4666 - val_acc: 0.8800\n",
"Epoch 98/10000\n",
"0s - loss: 0.4005 - acc: 0.8933 - val_loss: 0.4653 - val_acc: 0.8667\n",
"Epoch 99/10000\n",
"0s - loss: 0.4005 - acc: 0.9200 - val_loss: 0.4644 - val_acc: 0.9333\n",
"Epoch 100/10000\n",
"0s - loss: 0.3946 - acc: 0.9200 - val_loss: 0.4629 - val_acc: 0.8933\n",
"Epoch 101/10000\n",
"0s - loss: 0.3936 - acc: 0.9200 - val_loss: 0.4619 - val_acc: 0.9067\n",
"Epoch 102/10000\n",
"0s - loss: 0.3950 - acc: 0.8933 - val_loss: 0.4603 - val_acc: 0.8800\n",
"Epoch 103/10000\n",
"0s - loss: 0.3927 - acc: 0.9067 - val_loss: 0.4591 - val_acc: 0.8933\n",
"Epoch 104/10000\n",
"0s - loss: 0.3904 - acc: 0.9467 - val_loss: 0.4583 - val_acc: 0.9333\n",
"Epoch 105/10000\n",
"0s - loss: 0.3899 - acc: 0.9333 - val_loss: 0.4568 - val_acc: 0.8933\n",
"Epoch 106/10000\n",
"0s - loss: 0.3868 - acc: 0.9200 - val_loss: 0.4557 - val_acc: 0.9067\n",
"Epoch 107/10000\n",
"0s - loss: 0.3860 - acc: 0.9467 - val_loss: 0.4547 - val_acc: 0.9333\n",
"Epoch 108/10000\n",
"0s - loss: 0.3850 - acc: 0.9467 - val_loss: 0.4534 - val_acc: 0.9333\n",
"Epoch 109/10000\n",
"0s - loss: 0.3838 - acc: 0.9467 - val_loss: 0.4521 - val_acc: 0.9067\n",
"Epoch 110/10000\n",
"0s - loss: 0.3839 - acc: 0.9467 - val_loss: 0.4512 - val_acc: 0.9333\n",
"Epoch 111/10000\n",
"0s - loss: 0.3830 - acc: 0.9333 - val_loss: 0.4495 - val_acc: 0.8933\n",
"Epoch 112/10000\n",
"0s - loss: 0.3829 - acc: 0.9200 - val_loss: 0.4488 - val_acc: 0.8800\n",
"Epoch 113/10000\n",
"0s - loss: 0.3796 - acc: 0.9200 - val_loss: 0.4475 - val_acc: 0.9333\n",
"Epoch 114/10000\n",
"0s - loss: 0.3781 - acc: 0.9467 - val_loss: 0.4465 - val_acc: 0.9333\n",
"Epoch 115/10000\n",
"0s - loss: 0.3775 - acc: 0.9467 - val_loss: 0.4457 - val_acc: 0.9200\n",
"Epoch 116/10000\n",
"0s - loss: 0.3778 - acc: 0.9333 - val_loss: 0.4441 - val_acc: 0.9200\n",
"Epoch 117/10000\n",
"0s - loss: 0.3762 - acc: 0.9200 - val_loss: 0.4430 - val_acc: 0.9333\n",
"Epoch 118/10000\n",
"0s - loss: 0.3774 - acc: 0.9200 - val_loss: 0.4419 - val_acc: 0.8933\n",
"Epoch 119/10000\n",
"0s - loss: 0.3750 - acc: 0.9333 - val_loss: 0.4415 - val_acc: 0.9200\n",
"Epoch 120/10000\n",
"0s - loss: 0.3719 - acc: 0.9467 - val_loss: 0.4398 - val_acc: 0.9333\n",
"Epoch 121/10000\n",
"0s - loss: 0.3710 - acc: 0.9467 - val_loss: 0.4387 - val_acc: 0.9200\n",
"Epoch 122/10000\n",
"0s - loss: 0.3707 - acc: 0.9467 - val_loss: 0.4380 - val_acc: 0.9333\n",
"Epoch 123/10000\n",
"0s - loss: 0.3690 - acc: 0.9467 - val_loss: 0.4366 - val_acc: 0.9200\n",
"Epoch 124/10000\n",
"0s - loss: 0.3698 - acc: 0.9333 - val_loss: 0.4356 - val_acc: 0.9200\n",
"Epoch 125/10000\n",
"0s - loss: 0.3667 - acc: 0.9333 - val_loss: 0.4345 - val_acc: 0.9200\n",
"Epoch 126/10000\n",
"0s - loss: 0.3673 - acc: 0.9467 - val_loss: 0.4335 - val_acc: 0.9200\n",
"Epoch 127/10000\n",
"0s - loss: 0.3706 - acc: 0.9467 - val_loss: 0.4339 - val_acc: 0.9333\n",
"Epoch 128/10000\n",
"0s - loss: 0.3642 - acc: 0.9467 - val_loss: 0.4314 - val_acc: 0.9333\n",
"Epoch 129/10000\n",
"0s - loss: 0.3638 - acc: 0.9333 - val_loss: 0.4303 - val_acc: 0.9333\n",
"Epoch 130/10000\n",
"0s - loss: 0.3638 - acc: 0.9467 - val_loss: 0.4299 - val_acc: 0.9333\n",
"Epoch 131/10000\n",
"0s - loss: 0.3610 - acc: 0.9467 - val_loss: 0.4286 - val_acc: 0.9333\n",
"Epoch 132/10000\n",
"0s - loss: 0.3602 - acc: 0.9467 - val_loss: 0.4275 - val_acc: 0.9333\n",
"Epoch 133/10000\n",
"0s - loss: 0.3608 - acc: 0.9333 - val_loss: 0.4266 - val_acc: 0.9333\n",
"Epoch 134/10000\n",
"0s - loss: 0.3608 - acc: 0.9467 - val_loss: 0.4263 - val_acc: 0.9200\n",
"Epoch 135/10000\n",
"0s - loss: 0.3576 - acc: 0.9467 - val_loss: 0.4247 - val_acc: 0.9333\n",
"Epoch 136/10000\n",
"0s - loss: 0.3560 - acc: 0.9467 - val_loss: 0.4236 - val_acc: 0.9333\n",
"Epoch 137/10000\n",
"0s - loss: 0.3554 - acc: 0.9467 - val_loss: 0.4227 - val_acc: 0.9467\n",
"Epoch 138/10000\n",
"0s - loss: 0.3546 - acc: 0.9467 - val_loss: 0.4215 - val_acc: 0.9333\n",
"Epoch 139/10000\n",
"0s - loss: 0.3542 - acc: 0.9467 - val_loss: 0.4207 - val_acc: 0.9467\n",
"Epoch 140/10000\n",
"0s - loss: 0.3553 - acc: 0.9333 - val_loss: 0.4198 - val_acc: 0.9333\n",
"Epoch 141/10000\n",
"0s - loss: 0.3518 - acc: 0.9467 - val_loss: 0.4190 - val_acc: 0.9333\n",
"Epoch 142/10000\n",
"0s - loss: 0.3509 - acc: 0.9600 - val_loss: 0.4181 - val_acc: 0.9200\n",
"Epoch 143/10000\n",
"0s - loss: 0.3515 - acc: 0.9600 - val_loss: 0.4177 - val_acc: 0.9333\n",
"Epoch 144/10000\n",
"0s - loss: 0.3490 - acc: 0.9467 - val_loss: 0.4159 - val_acc: 0.9467\n",
"Epoch 145/10000\n",
"0s - loss: 0.3502 - acc: 0.9333 - val_loss: 0.4151 - val_acc: 0.9467\n",
"Epoch 146/10000\n",
"0s - loss: 0.3470 - acc: 0.9467 - val_loss: 0.4142 - val_acc: 0.9467\n",
"Epoch 147/10000\n",
"0s - loss: 0.3463 - acc: 0.9467 - val_loss: 0.4133 - val_acc: 0.9467\n",
"Epoch 148/10000\n",
"0s - loss: 0.3455 - acc: 0.9600 - val_loss: 0.4125 - val_acc: 0.9467\n",
"Epoch 149/10000\n",
"0s - loss: 0.3448 - acc: 0.9600 - val_loss: 0.4117 - val_acc: 0.9467\n",
"Epoch 150/10000\n",
"0s - loss: 0.3435 - acc: 0.9600 - val_loss: 0.4106 - val_acc: 0.9467\n",
"Epoch 151/10000\n",
"0s - loss: 0.3438 - acc: 0.9467 - val_loss: 0.4097 - val_acc: 0.9467\n",
"Epoch 152/10000\n",
"0s - loss: 0.3422 - acc: 0.9600 - val_loss: 0.4087 - val_acc: 0.9467\n",
"Epoch 153/10000\n",
"0s - loss: 0.3410 - acc: 0.9467 - val_loss: 0.4078 - val_acc: 0.9333\n",
"Epoch 154/10000\n",
"0s - loss: 0.3420 - acc: 0.9600 - val_loss: 0.4074 - val_acc: 0.9200\n",
"Epoch 155/10000\n",
"0s - loss: 0.3395 - acc: 0.9600 - val_loss: 0.4063 - val_acc: 0.9467\n",
"Epoch 156/10000\n",
"0s - loss: 0.3389 - acc: 0.9600 - val_loss: 0.4051 - val_acc: 0.9467\n",
"Epoch 157/10000\n",
"0s - loss: 0.3382 - acc: 0.9467 - val_loss: 0.4042 - val_acc: 0.9467\n",
"Epoch 158/10000\n",
"0s - loss: 0.3377 - acc: 0.9467 - val_loss: 0.4037 - val_acc: 0.9200\n",
"Epoch 159/10000\n",
"0s - loss: 0.3365 - acc: 0.9600 - val_loss: 0.4026 - val_acc: 0.9467\n",
"Epoch 160/10000\n",
"0s - loss: 0.3368 - acc: 0.9467 - val_loss: 0.4016 - val_acc: 0.9333\n",
"Epoch 161/10000\n",
"0s - loss: 0.3351 - acc: 0.9600 - val_loss: 0.4011 - val_acc: 0.9333\n",
"Epoch 162/10000\n",
"0s - loss: 0.3342 - acc: 0.9600 - val_loss: 0.4005 - val_acc: 0.9333\n",
"Epoch 163/10000\n",
"0s - loss: 0.3330 - acc: 0.9600 - val_loss: 0.3994 - val_acc: 0.9333\n",
"Epoch 164/10000\n",
"0s - loss: 0.3326 - acc: 0.9600 - val_loss: 0.3983 - val_acc: 0.9467\n",
"Epoch 165/10000\n",
"0s - loss: 0.3345 - acc: 0.9467 - val_loss: 0.3977 - val_acc: 0.9333\n",
"Epoch 166/10000\n",
"0s - loss: 0.3310 - acc: 0.9467 - val_loss: 0.3967 - val_acc: 0.9467\n",
"Epoch 167/10000\n",
"0s - loss: 0.3302 - acc: 0.9600 - val_loss: 0.3959 - val_acc: 0.9467\n",
"Epoch 168/10000\n",
"0s - loss: 0.3288 - acc: 0.9600 - val_loss: 0.3955 - val_acc: 0.9333\n",
"Epoch 169/10000\n",
"0s - loss: 0.3293 - acc: 0.9600 - val_loss: 0.3949 - val_acc: 0.9333\n",
"Epoch 170/10000\n",
"0s - loss: 0.3307 - acc: 0.9600 - val_loss: 0.3932 - val_acc: 0.9333\n",
"Epoch 171/10000\n",
"0s - loss: 0.3303 - acc: 0.9600 - val_loss: 0.3941 - val_acc: 0.9333\n",
"Epoch 172/10000\n",
"0s - loss: 0.3278 - acc: 0.9600 - val_loss: 0.3918 - val_acc: 0.9600\n",
"Epoch 173/10000\n",
"0s - loss: 0.3262 - acc: 0.9600 - val_loss: 0.3914 - val_acc: 0.9333\n",
"Epoch 174/10000\n",
"0s - loss: 0.3252 - acc: 0.9600 - val_loss: 0.3906 - val_acc: 0.9333\n",
"Epoch 175/10000\n",
"0s - loss: 0.3247 - acc: 0.9600 - val_loss: 0.3894 - val_acc: 0.9467\n",
"Epoch 176/10000\n",
"0s - loss: 0.3233 - acc: 0.9600 - val_loss: 0.3885 - val_acc: 0.9467\n",
"Epoch 177/10000\n",
"0s - loss: 0.3222 - acc: 0.9600 - val_loss: 0.3878 - val_acc: 0.9600\n",
"Epoch 178/10000\n",
"0s - loss: 0.3220 - acc: 0.9600 - val_loss: 0.3870 - val_acc: 0.9467\n",
"Epoch 179/10000\n",
"0s - loss: 0.3224 - acc: 0.9600 - val_loss: 0.3861 - val_acc: 0.9600\n",
"Epoch 180/10000\n",
"0s - loss: 0.3208 - acc: 0.9600 - val_loss: 0.3853 - val_acc: 0.9467\n",
"Epoch 181/10000\n",
"0s - loss: 0.3205 - acc: 0.9600 - val_loss: 0.3854 - val_acc: 0.9333\n",
"Epoch 182/10000\n",
"0s - loss: 0.3196 - acc: 0.9600 - val_loss: 0.3843 - val_acc: 0.9333\n",
"Epoch 183/10000\n",
"0s - loss: 0.3186 - acc: 0.9600 - val_loss: 0.3831 - val_acc: 0.9600\n",
"Epoch 184/10000\n",
"0s - loss: 0.3190 - acc: 0.9600 - val_loss: 0.3822 - val_acc: 0.9467\n",
"Epoch 185/10000\n",
"0s - loss: 0.3166 - acc: 0.9600 - val_loss: 0.3816 - val_acc: 0.9467\n",
"Epoch 186/10000\n",
"0s - loss: 0.3158 - acc: 0.9600 - val_loss: 0.3809 - val_acc: 0.9467\n",
"Epoch 187/10000\n",
"0s - loss: 0.3162 - acc: 0.9600 - val_loss: 0.3810 - val_acc: 0.9467\n",
"Epoch 188/10000\n",
"0s - loss: 0.3147 - acc: 0.9600 - val_loss: 0.3793 - val_acc: 0.9467\n",
"Epoch 189/10000\n",
"0s - loss: 0.3148 - acc: 0.9600 - val_loss: 0.3792 - val_acc: 0.9333\n",
"Epoch 190/10000\n",
"0s - loss: 0.3136 - acc: 0.9600 - val_loss: 0.3776 - val_acc: 0.9600\n",
"Epoch 191/10000\n",
"0s - loss: 0.3126 - acc: 0.9600 - val_loss: 0.3769 - val_acc: 0.9600\n",
"Epoch 192/10000\n",
"0s - loss: 0.3123 - acc: 0.9600 - val_loss: 0.3769 - val_acc: 0.9333\n",
"Epoch 193/10000\n",
"0s - loss: 0.3119 - acc: 0.9600 - val_loss: 0.3763 - val_acc: 0.9333\n",
"Epoch 194/10000\n",
"0s - loss: 0.3109 - acc: 0.9600 - val_loss: 0.3746 - val_acc: 0.9600\n",
"Epoch 195/10000\n",
"0s - loss: 0.3094 - acc: 0.9600 - val_loss: 0.3741 - val_acc: 0.9467\n",
"Epoch 196/10000\n",
"0s - loss: 0.3116 - acc: 0.9600 - val_loss: 0.3740 - val_acc: 0.9333\n",
"Epoch 197/10000\n",
"0s - loss: 0.3088 - acc: 0.9600 - val_loss: 0.3727 - val_acc: 0.9467\n",
"Epoch 198/10000\n",
"0s - loss: 0.3082 - acc: 0.9600 - val_loss: 0.3721 - val_acc: 0.9467\n",
"Epoch 199/10000\n",
"0s - loss: 0.3066 - acc: 0.9600 - val_loss: 0.3710 - val_acc: 0.9467\n",
"Epoch 200/10000\n",
"0s - loss: 0.3064 - acc: 0.9600 - val_loss: 0.3702 - val_acc: 0.9600\n",
"Epoch 201/10000\n",
"0s - loss: 0.3082 - acc: 0.9600 - val_loss: 0.3700 - val_acc: 0.9467\n",
"Epoch 202/10000\n",
"0s - loss: 0.3071 - acc: 0.9600 - val_loss: 0.3693 - val_acc: 0.9467\n",
"Epoch 203/10000\n",
"0s - loss: 0.3044 - acc: 0.9600 - val_loss: 0.3684 - val_acc: 0.9467\n",
"Epoch 204/10000\n",
"0s - loss: 0.3039 - acc: 0.9600 - val_loss: 0.3678 - val_acc: 0.9467\n",
"Epoch 205/10000\n",
"0s - loss: 0.3043 - acc: 0.9600 - val_loss: 0.3667 - val_acc: 0.9600\n",
"Epoch 206/10000\n",
"0s - loss: 0.3027 - acc: 0.9600 - val_loss: 0.3659 - val_acc: 0.9600\n",
"Epoch 207/10000\n",
"0s - loss: 0.3021 - acc: 0.9600 - val_loss: 0.3653 - val_acc: 0.9467\n",
"Epoch 208/10000\n",
"0s - loss: 0.3011 - acc: 0.9600 - val_loss: 0.3651 - val_acc: 0.9600\n",
"Epoch 209/10000\n",
"0s - loss: 0.3017 - acc: 0.9600 - val_loss: 0.3649 - val_acc: 0.9467\n",
"Epoch 210/10000\n",
"0s - loss: 0.2999 - acc: 0.9600 - val_loss: 0.3635 - val_acc: 0.9467\n",
"Epoch 211/10000\n",
"0s - loss: 0.2992 - acc: 0.9600 - val_loss: 0.3626 - val_acc: 0.9467\n",
"Epoch 212/10000\n",
"0s - loss: 0.2985 - acc: 0.9600 - val_loss: 0.3624 - val_acc: 0.9600\n",
"Epoch 213/10000\n",
"0s - loss: 0.2987 - acc: 0.9600 - val_loss: 0.3623 - val_acc: 0.9467\n",
"Epoch 214/10000\n",
"0s - loss: 0.2974 - acc: 0.9600 - val_loss: 0.3613 - val_acc: 0.9600\n",
"Epoch 215/10000\n",
"0s - loss: 0.2962 - acc: 0.9600 - val_loss: 0.3598 - val_acc: 0.9467\n",
"Epoch 216/10000\n",
"0s - loss: 0.2969 - acc: 0.9600 - val_loss: 0.3590 - val_acc: 0.9600\n",
"Epoch 217/10000\n",
"0s - loss: 0.2955 - acc: 0.9600 - val_loss: 0.3584 - val_acc: 0.9467\n",
"Epoch 218/10000\n",
"0s - loss: 0.2948 - acc: 0.9600 - val_loss: 0.3580 - val_acc: 0.9600\n",
"Epoch 219/10000\n",
"0s - loss: 0.2949 - acc: 0.9600 - val_loss: 0.3577 - val_acc: 0.9600\n",
"Epoch 220/10000\n",
"0s - loss: 0.2933 - acc: 0.9600 - val_loss: 0.3564 - val_acc: 0.9467\n",
"Epoch 221/10000\n",
"0s - loss: 0.2926 - acc: 0.9600 - val_loss: 0.3559 - val_acc: 0.9600\n",
"Epoch 222/10000\n",
"0s - loss: 0.2933 - acc: 0.9600 - val_loss: 0.3551 - val_acc: 0.9600\n",
"Epoch 223/10000\n",
"0s - loss: 0.2919 - acc: 0.9600 - val_loss: 0.3547 - val_acc: 0.9600\n",
"Epoch 224/10000\n",
"0s - loss: 0.2930 - acc: 0.9600 - val_loss: 0.3547 - val_acc: 0.9600\n",
"Epoch 225/10000\n",
"0s - loss: 0.2910 - acc: 0.9600 - val_loss: 0.3536 - val_acc: 0.9600\n",
"Epoch 226/10000\n",
"0s - loss: 0.2896 - acc: 0.9600 - val_loss: 0.3527 - val_acc: 0.9600\n",
"Epoch 227/10000\n",
"0s - loss: 0.2904 - acc: 0.9600 - val_loss: 0.3519 - val_acc: 0.9467\n",
"Epoch 228/10000\n",
"0s - loss: 0.2883 - acc: 0.9600 - val_loss: 0.3514 - val_acc: 0.9600\n",
"Epoch 229/10000\n",
"0s - loss: 0.2881 - acc: 0.9600 - val_loss: 0.3509 - val_acc: 0.9600\n",
"Epoch 230/10000\n",
"0s - loss: 0.2880 - acc: 0.9600 - val_loss: 0.3502 - val_acc: 0.9600\n",
"Epoch 231/10000\n",
"0s - loss: 0.2869 - acc: 0.9600 - val_loss: 0.3501 - val_acc: 0.9600\n",
"Epoch 232/10000\n",
"0s - loss: 0.2864 - acc: 0.9600 - val_loss: 0.3495 - val_acc: 0.9600\n",
"Epoch 233/10000\n",
"0s - loss: 0.2870 - acc: 0.9600 - val_loss: 0.3482 - val_acc: 0.9600\n",
"Epoch 234/10000\n",
"0s - loss: 0.2852 - acc: 0.9600 - val_loss: 0.3481 - val_acc: 0.9600\n",
"Epoch 235/10000\n",
"0s - loss: 0.2861 - acc: 0.9600 - val_loss: 0.3481 - val_acc: 0.9467\n",
"Epoch 236/10000\n",
"0s - loss: 0.2846 - acc: 0.9600 - val_loss: 0.3461 - val_acc: 0.9600\n",
"Epoch 237/10000\n",
"0s - loss: 0.2836 - acc: 0.9600 - val_loss: 0.3455 - val_acc: 0.9600\n",
"Epoch 238/10000\n",
"0s - loss: 0.2828 - acc: 0.9600 - val_loss: 0.3449 - val_acc: 0.9600\n",
"Epoch 239/10000\n",
"0s - loss: 0.2821 - acc: 0.9600 - val_loss: 0.3446 - val_acc: 0.9600\n",
"Epoch 240/10000\n",
"0s - loss: 0.2827 - acc: 0.9600 - val_loss: 0.3437 - val_acc: 0.9600\n",
"Epoch 241/10000\n",
"0s - loss: 0.2819 - acc: 0.9600 - val_loss: 0.3433 - val_acc: 0.9600\n",
"Epoch 242/10000\n",
"0s - loss: 0.2807 - acc: 0.9600 - val_loss: 0.3423 - val_acc: 0.9600\n",
"Epoch 243/10000\n",
"0s - loss: 0.2808 - acc: 0.9600 - val_loss: 0.3419 - val_acc: 0.9600\n",
"Epoch 244/10000\n",
"0s - loss: 0.2793 - acc: 0.9600 - val_loss: 0.3414 - val_acc: 0.9600\n",
"Epoch 245/10000\n",
"0s - loss: 0.2796 - acc: 0.9600 - val_loss: 0.3403 - val_acc: 0.9600\n",
"Epoch 246/10000\n",
"0s - loss: 0.2793 - acc: 0.9600 - val_loss: 0.3404 - val_acc: 0.9600\n",
"Epoch 247/10000\n",
"0s - loss: 0.2825 - acc: 0.9600 - val_loss: 0.3391 - val_acc: 0.9733\n",
"Epoch 248/10000\n",
"0s - loss: 0.2780 - acc: 0.9600 - val_loss: 0.3393 - val_acc: 0.9600\n",
"Epoch 249/10000\n",
"0s - loss: 0.2766 - acc: 0.9600 - val_loss: 0.3389 - val_acc: 0.9600\n",
"Epoch 250/10000\n",
"0s - loss: 0.2763 - acc: 0.9600 - val_loss: 0.3378 - val_acc: 0.9600\n",
"Epoch 251/10000\n",
"0s - loss: 0.2768 - acc: 0.9600 - val_loss: 0.3379 - val_acc: 0.9600\n",
"Epoch 252/10000\n",
"0s - loss: 0.2750 - acc: 0.9733 - val_loss: 0.3370 - val_acc: 0.9600\n",
"Epoch 253/10000\n",
"0s - loss: 0.2745 - acc: 0.9600 - val_loss: 0.3362 - val_acc: 0.9600\n",
"Epoch 254/10000\n",
"0s - loss: 0.2738 - acc: 0.9600 - val_loss: 0.3349 - val_acc: 0.9600\n",
"Epoch 255/10000\n",
"0s - loss: 0.2731 - acc: 0.9600 - val_loss: 0.3344 - val_acc: 0.9600\n",
"Epoch 256/10000\n",
"0s - loss: 0.2733 - acc: 0.9600 - val_loss: 0.3340 - val_acc: 0.9600\n",
"Epoch 257/10000\n",
"0s - loss: 0.2725 - acc: 0.9600 - val_loss: 0.3337 - val_acc: 0.9600\n",
"Epoch 258/10000\n",
"0s - loss: 0.2721 - acc: 0.9733 - val_loss: 0.3336 - val_acc: 0.9600\n",
"Epoch 259/10000\n",
"0s - loss: 0.2736 - acc: 0.9600 - val_loss: 0.3318 - val_acc: 0.9600\n",
"Epoch 260/10000\n",
"0s - loss: 0.2703 - acc: 0.9600 - val_loss: 0.3317 - val_acc: 0.9600\n",
"Epoch 261/10000\n",
"0s - loss: 0.2700 - acc: 0.9600 - val_loss: 0.3311 - val_acc: 0.9600\n",
"Epoch 262/10000\n",
"0s - loss: 0.2704 - acc: 0.9733 - val_loss: 0.3312 - val_acc: 0.9600\n",
"Epoch 263/10000\n",
"0s - loss: 0.2689 - acc: 0.9600 - val_loss: 0.3300 - val_acc: 0.9600\n",
"Epoch 264/10000\n",
"0s - loss: 0.2690 - acc: 0.9733 - val_loss: 0.3300 - val_acc: 0.9600\n",
"Epoch 265/10000\n",
"0s - loss: 0.2681 - acc: 0.9733 - val_loss: 0.3292 - val_acc: 0.9600\n",
"Epoch 266/10000\n",
"0s - loss: 0.2672 - acc: 0.9600 - val_loss: 0.3282 - val_acc: 0.9600\n",
"Epoch 267/10000\n",
"0s - loss: 0.2669 - acc: 0.9600 - val_loss: 0.3274 - val_acc: 0.9600\n",
"Epoch 268/10000\n",
"0s - loss: 0.2664 - acc: 0.9600 - val_loss: 0.3269 - val_acc: 0.9600\n",
"Epoch 269/10000\n",
"0s - loss: 0.2662 - acc: 0.9600 - val_loss: 0.3263 - val_acc: 0.9600\n",
"Epoch 270/10000\n",
"0s - loss: 0.2661 - acc: 0.9600 - val_loss: 0.3256 - val_acc: 0.9600\n",
"Epoch 271/10000\n",
"0s - loss: 0.2653 - acc: 0.9600 - val_loss: 0.3251 - val_acc: 0.9600\n",
"Epoch 272/10000\n",
"0s - loss: 0.2643 - acc: 0.9600 - val_loss: 0.3252 - val_acc: 0.9600\n",
"Epoch 273/10000\n",
"0s - loss: 0.2643 - acc: 0.9600 - val_loss: 0.3242 - val_acc: 0.9600\n",
"Epoch 274/10000\n",
"0s - loss: 0.2649 - acc: 0.9600 - val_loss: 0.3234 - val_acc: 0.9600\n",
"Epoch 275/10000\n",
"0s - loss: 0.2650 - acc: 0.9600 - val_loss: 0.3248 - val_acc: 0.9600\n",
"Epoch 276/10000\n",
"0s - loss: 0.2628 - acc: 0.9600 - val_loss: 0.3225 - val_acc: 0.9600\n",
"Epoch 277/10000\n",
"0s - loss: 0.2622 - acc: 0.9733 - val_loss: 0.3220 - val_acc: 0.9600\n",
"Epoch 278/10000\n",
"0s - loss: 0.2619 - acc: 0.9733 - val_loss: 0.3216 - val_acc: 0.9600\n",
"Epoch 279/10000\n",
"0s - loss: 0.2611 - acc: 0.9733 - val_loss: 0.3214 - val_acc: 0.9600\n",
"Epoch 280/10000\n",
"0s - loss: 0.2604 - acc: 0.9733 - val_loss: 0.3207 - val_acc: 0.9600\n",
"Epoch 281/10000\n",
"0s - loss: 0.2595 - acc: 0.9600 - val_loss: 0.3197 - val_acc: 0.9600\n",
"Epoch 282/10000\n",
"0s - loss: 0.2595 - acc: 0.9600 - val_loss: 0.3192 - val_acc: 0.9600\n",
"Epoch 283/10000\n",
"0s - loss: 0.2607 - acc: 0.9600 - val_loss: 0.3181 - val_acc: 0.9600\n",
"Epoch 284/10000\n",
"0s - loss: 0.2587 - acc: 0.9600 - val_loss: 0.3179 - val_acc: 0.9600\n",
"Epoch 285/10000\n",
"0s - loss: 0.2581 - acc: 0.9733 - val_loss: 0.3178 - val_acc: 0.9600\n",
"Epoch 286/10000\n",
"0s - loss: 0.2572 - acc: 0.9733 - val_loss: 0.3170 - val_acc: 0.9600\n",
"Epoch 287/10000\n",
"0s - loss: 0.2571 - acc: 0.9733 - val_loss: 0.3165 - val_acc: 0.9600\n",
"Epoch 288/10000\n",
"0s - loss: 0.2571 - acc: 0.9733 - val_loss: 0.3173 - val_acc: 0.9600\n",
"Epoch 289/10000\n",
"0s - loss: 0.2562 - acc: 0.9733 - val_loss: 0.3160 - val_acc: 0.9600\n",
"Epoch 290/10000\n",
"0s - loss: 0.2556 - acc: 0.9600 - val_loss: 0.3146 - val_acc: 0.9600\n",
"Epoch 291/10000\n",
"0s - loss: 0.2548 - acc: 0.9600 - val_loss: 0.3144 - val_acc: 0.9600\n",
"Epoch 292/10000\n",
"0s - loss: 0.2544 - acc: 0.9733 - val_loss: 0.3140 - val_acc: 0.9600\n",
"Epoch 293/10000\n",
"0s - loss: 0.2551 - acc: 0.9733 - val_loss: 0.3136 - val_acc: 0.9600\n",
"Epoch 294/10000\n",
"0s - loss: 0.2531 - acc: 0.9733 - val_loss: 0.3123 - val_acc: 0.9600\n",
"Epoch 295/10000\n",
"0s - loss: 0.2533 - acc: 0.9600 - val_loss: 0.3118 - val_acc: 0.9600\n",
"Epoch 296/10000\n",
"0s - loss: 0.2524 - acc: 0.9600 - val_loss: 0.3112 - val_acc: 0.9600\n",
"Epoch 297/10000\n",
"0s - loss: 0.2516 - acc: 0.9733 - val_loss: 0.3114 - val_acc: 0.9600\n",
"Epoch 298/10000\n",
"0s - loss: 0.2515 - acc: 0.9733 - val_loss: 0.3112 - val_acc: 0.9600\n",
"Epoch 299/10000\n",
"0s - loss: 0.2513 - acc: 0.9733 - val_loss: 0.3108 - val_acc: 0.9600\n",
"Epoch 300/10000\n",
"0s - loss: 0.2507 - acc: 0.9733 - val_loss: 0.3098 - val_acc: 0.9600\n",
"Epoch 301/10000\n",
"0s - loss: 0.2507 - acc: 0.9733 - val_loss: 0.3094 - val_acc: 0.9600\n",
"Epoch 302/10000\n",
"0s - loss: 0.2498 - acc: 0.9733 - val_loss: 0.3082 - val_acc: 0.9600\n",
"Epoch 303/10000\n",
"0s - loss: 0.2497 - acc: 0.9733 - val_loss: 0.3080 - val_acc: 0.9600\n",
"Epoch 304/10000\n",
"0s - loss: 0.2487 - acc: 0.9733 - val_loss: 0.3075 - val_acc: 0.9600\n",
"Epoch 305/10000\n",
"0s - loss: 0.2482 - acc: 0.9733 - val_loss: 0.3070 - val_acc: 0.9600\n",
"Epoch 306/10000\n",
"0s - loss: 0.2482 - acc: 0.9600 - val_loss: 0.3060 - val_acc: 0.9600\n",
"Epoch 307/10000\n",
"0s - loss: 0.2476 - acc: 0.9600 - val_loss: 0.3055 - val_acc: 0.9600\n",
"Epoch 308/10000\n",
"0s - loss: 0.2485 - acc: 0.9733 - val_loss: 0.3068 - val_acc: 0.9600\n",
"Epoch 309/10000\n",
"0s - loss: 0.2460 - acc: 0.9733 - val_loss: 0.3050 - val_acc: 0.9600\n",
"Epoch 310/10000\n",
"0s - loss: 0.2464 - acc: 0.9733 - val_loss: 0.3041 - val_acc: 0.9600\n",
"Epoch 311/10000\n",
"0s - loss: 0.2465 - acc: 0.9733 - val_loss: 0.3041 - val_acc: 0.9600\n",
"Epoch 312/10000\n",
"0s - loss: 0.2450 - acc: 0.9733 - val_loss: 0.3039 - val_acc: 0.9600\n",
"Epoch 313/10000\n",
"0s - loss: 0.2447 - acc: 0.9733 - val_loss: 0.3029 - val_acc: 0.9600\n",
"Epoch 314/10000\n",
"0s - loss: 0.2440 - acc: 0.9733 - val_loss: 0.3024 - val_acc: 0.9600\n",
"Epoch 315/10000\n",
"0s - loss: 0.2438 - acc: 0.9733 - val_loss: 0.3018 - val_acc: 0.9600\n",
"Epoch 316/10000\n",
"0s - loss: 0.2439 - acc: 0.9600 - val_loss: 0.3010 - val_acc: 0.9600\n",
"Epoch 317/10000\n",
"0s - loss: 0.2428 - acc: 0.9733 - val_loss: 0.3009 - val_acc: 0.9600\n",
"Epoch 318/10000\n",
"0s - loss: 0.2438 - acc: 0.9733 - val_loss: 0.3013 - val_acc: 0.9600\n",
"Epoch 319/10000\n",
"0s - loss: 0.2424 - acc: 0.9733 - val_loss: 0.2996 - val_acc: 0.9600\n",
"Epoch 320/10000\n",
"0s - loss: 0.2432 - acc: 0.9733 - val_loss: 0.3001 - val_acc: 0.9600\n",
"Epoch 321/10000\n",
"0s - loss: 0.2421 - acc: 0.9733 - val_loss: 0.2996 - val_acc: 0.9600\n",
"Epoch 322/10000\n",
"0s - loss: 0.2412 - acc: 0.9733 - val_loss: 0.2981 - val_acc: 0.9600\n",
"Epoch 323/10000\n",
"0s - loss: 0.2399 - acc: 0.9733 - val_loss: 0.2978 - val_acc: 0.9600\n",
"Epoch 324/10000\n",
"0s - loss: 0.2414 - acc: 0.9733 - val_loss: 0.2979 - val_acc: 0.9600\n",
"Epoch 325/10000\n",
"0s - loss: 0.2419 - acc: 0.9600 - val_loss: 0.2965 - val_acc: 0.9600\n",
"Epoch 326/10000\n",
"0s - loss: 0.2388 - acc: 0.9733 - val_loss: 0.2972 - val_acc: 0.9600\n",
"Epoch 327/10000\n",
"0s - loss: 0.2401 - acc: 0.9733 - val_loss: 0.2956 - val_acc: 0.9600\n",
"Epoch 328/10000\n",
"0s - loss: 0.2378 - acc: 0.9733 - val_loss: 0.2960 - val_acc: 0.9600\n",
"Epoch 329/10000\n",
"0s - loss: 0.2386 - acc: 0.9733 - val_loss: 0.2953 - val_acc: 0.9600\n",
"Epoch 330/10000\n",
"0s - loss: 0.2376 - acc: 0.9733 - val_loss: 0.2952 - val_acc: 0.9600\n",
"Epoch 331/10000\n",
"0s - loss: 0.2374 - acc: 0.9733 - val_loss: 0.2936 - val_acc: 0.9600\n",
"Epoch 332/10000\n",
"0s - loss: 0.2372 - acc: 0.9733 - val_loss: 0.2946 - val_acc: 0.9600\n",
"Epoch 333/10000\n",
"0s - loss: 0.2354 - acc: 0.9733 - val_loss: 0.2935 - val_acc: 0.9600\n",
"Epoch 334/10000\n",
"0s - loss: 0.2354 - acc: 0.9733 - val_loss: 0.2934 - val_acc: 0.9600\n",
"Epoch 335/10000\n",
"0s - loss: 0.2368 - acc: 0.9733 - val_loss: 0.2916 - val_acc: 0.9600\n",
"Epoch 336/10000\n",
"0s - loss: 0.2340 - acc: 0.9733 - val_loss: 0.2915 - val_acc: 0.9600\n",
"Epoch 337/10000\n",
"0s - loss: 0.2338 - acc: 0.9733 - val_loss: 0.2920 - val_acc: 0.9600\n",
"Epoch 338/10000\n",
"0s - loss: 0.2339 - acc: 0.9733 - val_loss: 0.2918 - val_acc: 0.9600\n",
"Epoch 339/10000\n",
"0s - loss: 0.2336 - acc: 0.9733 - val_loss: 0.2909 - val_acc: 0.9600\n",
"Epoch 340/10000\n",
"0s - loss: 0.2331 - acc: 0.9733 - val_loss: 0.2897 - val_acc: 0.9600\n",
"Epoch 341/10000\n",
"0s - loss: 0.2331 - acc: 0.9733 - val_loss: 0.2892 - val_acc: 0.9600\n",
"Epoch 342/10000\n",
"0s - loss: 0.2323 - acc: 0.9733 - val_loss: 0.2890 - val_acc: 0.9600\n",
"Epoch 343/10000\n",
"0s - loss: 0.2319 - acc: 0.9733 - val_loss: 0.2894 - val_acc: 0.9600\n",
"Epoch 344/10000\n",
"0s - loss: 0.2322 - acc: 0.9733 - val_loss: 0.2878 - val_acc: 0.9600\n",
"Epoch 345/10000\n",
"0s - loss: 0.2306 - acc: 0.9733 - val_loss: 0.2873 - val_acc: 0.9600\n",
"Epoch 346/10000\n",
"0s - loss: 0.2312 - acc: 0.9733 - val_loss: 0.2868 - val_acc: 0.9600\n",
"Epoch 347/10000\n",
"0s - loss: 0.2305 - acc: 0.9733 - val_loss: 0.2878 - val_acc: 0.9600\n",
"Epoch 348/10000\n",
"0s - loss: 0.2292 - acc: 0.9733 - val_loss: 0.2868 - val_acc: 0.9600\n",
"Epoch 349/10000\n",
"0s - loss: 0.2288 - acc: 0.9733 - val_loss: 0.2861 - val_acc: 0.9600\n",
"Epoch 350/10000\n",
"0s - loss: 0.2285 - acc: 0.9733 - val_loss: 0.2852 - val_acc: 0.9600\n",
"Epoch 351/10000\n",
"0s - loss: 0.2293 - acc: 0.9733 - val_loss: 0.2843 - val_acc: 0.9600\n",
"Epoch 352/10000\n",
"0s - loss: 0.2280 - acc: 0.9733 - val_loss: 0.2845 - val_acc: 0.9600\n",
"Epoch 353/10000\n",
"0s - loss: 0.2283 - acc: 0.9733 - val_loss: 0.2837 - val_acc: 0.9600\n",
"Epoch 354/10000\n",
"0s - loss: 0.2266 - acc: 0.9733 - val_loss: 0.2843 - val_acc: 0.9600\n",
"Epoch 355/10000\n",
"0s - loss: 0.2270 - acc: 0.9733 - val_loss: 0.2833 - val_acc: 0.9600\n",
"Epoch 356/10000\n",
"0s - loss: 0.2264 - acc: 0.9733 - val_loss: 0.2840 - val_acc: 0.9733\n",
"Epoch 357/10000\n",
"0s - loss: 0.2259 - acc: 0.9733 - val_loss: 0.2832 - val_acc: 0.9600\n",
"Epoch 358/10000\n",
"0s - loss: 0.2263 - acc: 0.9733 - val_loss: 0.2814 - val_acc: 0.9600\n",
"Epoch 359/10000\n",
"0s - loss: 0.2250 - acc: 0.9733 - val_loss: 0.2816 - val_acc: 0.9600\n",
"Epoch 360/10000\n",
"0s - loss: 0.2255 - acc: 0.9733 - val_loss: 0.2815 - val_acc: 0.9600\n",
"Epoch 361/10000\n",
"0s - loss: 0.2244 - acc: 0.9733 - val_loss: 0.2809 - val_acc: 0.9600\n",
"Epoch 362/10000\n",
"0s - loss: 0.2238 - acc: 0.9733 - val_loss: 0.2801 - val_acc: 0.9600\n",
"Epoch 363/10000\n",
"0s - loss: 0.2237 - acc: 0.9733 - val_loss: 0.2801 - val_acc: 0.9600\n",
"Epoch 364/10000\n",
"0s - loss: 0.2232 - acc: 0.9733 - val_loss: 0.2788 - val_acc: 0.9600\n",
"Epoch 365/10000\n",
"0s - loss: 0.2234 - acc: 0.9733 - val_loss: 0.2785 - val_acc: 0.9600\n",
"Epoch 366/10000\n",
"0s - loss: 0.2239 - acc: 0.9733 - val_loss: 0.2786 - val_acc: 0.9600\n",
"Epoch 367/10000\n",
"0s - loss: 0.2221 - acc: 0.9733 - val_loss: 0.2775 - val_acc: 0.9600\n",
"Epoch 368/10000\n",
"0s - loss: 0.2233 - acc: 0.9733 - val_loss: 0.2787 - val_acc: 0.9733\n",
"Epoch 369/10000\n",
"0s - loss: 0.2215 - acc: 0.9733 - val_loss: 0.2768 - val_acc: 0.9600\n",
"Epoch 370/10000\n",
"0s - loss: 0.2207 - acc: 0.9733 - val_loss: 0.2763 - val_acc: 0.9600\n",
"Epoch 371/10000\n",
"0s - loss: 0.2203 - acc: 0.9733 - val_loss: 0.2760 - val_acc: 0.9600\n",
"Epoch 372/10000\n",
"0s - loss: 0.2203 - acc: 0.9733 - val_loss: 0.2771 - val_acc: 0.9733\n",
"Epoch 373/10000\n",
"0s - loss: 0.2198 - acc: 0.9733 - val_loss: 0.2757 - val_acc: 0.9600\n",
"Epoch 374/10000\n",
"0s - loss: 0.2197 - acc: 0.9733 - val_loss: 0.2752 - val_acc: 0.9600\n",
"Epoch 375/10000\n",
"0s - loss: 0.2192 - acc: 0.9733 - val_loss: 0.2750 - val_acc: 0.9733\n",
"Epoch 376/10000\n",
"0s - loss: 0.2189 - acc: 0.9733 - val_loss: 0.2742 - val_acc: 0.9600\n",
"Epoch 377/10000\n",
"0s - loss: 0.2183 - acc: 0.9733 - val_loss: 0.2732 - val_acc: 0.9600\n",
"Epoch 378/10000\n",
"0s - loss: 0.2179 - acc: 0.9733 - val_loss: 0.2736 - val_acc: 0.9600\n",
"Epoch 379/10000\n",
"0s - loss: 0.2195 - acc: 0.9733 - val_loss: 0.2739 - val_acc: 0.9733\n",
"Epoch 380/10000\n",
"0s - loss: 0.2168 - acc: 0.9733 - val_loss: 0.2722 - val_acc: 0.9600\n",
"Epoch 381/10000\n",
"0s - loss: 0.2170 - acc: 0.9733 - val_loss: 0.2715 - val_acc: 0.9600\n",
"Epoch 382/10000\n",
"0s - loss: 0.2165 - acc: 0.9733 - val_loss: 0.2713 - val_acc: 0.9600\n",
"Epoch 383/10000\n",
"0s - loss: 0.2162 - acc: 0.9733 - val_loss: 0.2715 - val_acc: 0.9733\n",
"Epoch 384/10000\n",
"0s - loss: 0.2154 - acc: 0.9733 - val_loss: 0.2713 - val_acc: 0.9733\n",
"Epoch 385/10000\n",
"0s - loss: 0.2156 - acc: 0.9733 - val_loss: 0.2703 - val_acc: 0.9600\n",
"Epoch 386/10000\n",
"0s - loss: 0.2147 - acc: 0.9733 - val_loss: 0.2700 - val_acc: 0.9600\n",
"Epoch 387/10000\n",
"0s - loss: 0.2146 - acc: 0.9733 - val_loss: 0.2703 - val_acc: 0.9733\n",
"Epoch 388/10000\n",
"0s - loss: 0.2143 - acc: 0.9733 - val_loss: 0.2694 - val_acc: 0.9733\n",
"Epoch 389/10000\n",
"0s - loss: 0.2136 - acc: 0.9733 - val_loss: 0.2691 - val_acc: 0.9733\n",
"Epoch 390/10000\n",
"0s - loss: 0.2132 - acc: 0.9733 - val_loss: 0.2690 - val_acc: 0.9733\n",
"Epoch 391/10000\n",
"0s - loss: 0.2131 - acc: 0.9733 - val_loss: 0.2687 - val_acc: 0.9733\n",
"Epoch 392/10000\n",
"0s - loss: 0.2133 - acc: 0.9733 - val_loss: 0.2668 - val_acc: 0.9600\n",
"Epoch 393/10000\n",
"0s - loss: 0.2121 - acc: 0.9733 - val_loss: 0.2670 - val_acc: 0.9600\n",
"Epoch 394/10000\n",
"0s - loss: 0.2121 - acc: 0.9733 - val_loss: 0.2666 - val_acc: 0.9600\n",
"Epoch 395/10000\n",
"0s - loss: 0.2130 - acc: 0.9733 - val_loss: 0.2664 - val_acc: 0.9600\n",
"Epoch 396/10000\n",
"0s - loss: 0.2113 - acc: 0.9733 - val_loss: 0.2670 - val_acc: 0.9733\n",
"Epoch 397/10000\n",
"0s - loss: 0.2114 - acc: 0.9733 - val_loss: 0.2664 - val_acc: 0.9733\n",
"Epoch 398/10000\n",
"0s - loss: 0.2118 - acc: 0.9733 - val_loss: 0.2659 - val_acc: 0.9733\n",
"Epoch 399/10000\n",
"0s - loss: 0.2104 - acc: 0.9733 - val_loss: 0.2644 - val_acc: 0.9600\n",
"Epoch 400/10000\n",
"0s - loss: 0.2110 - acc: 0.9733 - val_loss: 0.2656 - val_acc: 0.9733\n",
"Epoch 401/10000\n",
"0s - loss: 0.2094 - acc: 0.9733 - val_loss: 0.2638 - val_acc: 0.9600\n",
"Epoch 402/10000\n",
"0s - loss: 0.2091 - acc: 0.9733 - val_loss: 0.2638 - val_acc: 0.9733\n",
"Epoch 403/10000\n",
"0s - loss: 0.2099 - acc: 0.9733 - val_loss: 0.2628 - val_acc: 0.9600\n",
"Epoch 404/10000\n",
"0s - loss: 0.2087 - acc: 0.9733 - val_loss: 0.2622 - val_acc: 0.9600\n",
"Epoch 405/10000\n",
"0s - loss: 0.2086 - acc: 0.9733 - val_loss: 0.2626 - val_acc: 0.9733\n",
"Epoch 406/10000\n",
"0s - loss: 0.2080 - acc: 0.9733 - val_loss: 0.2619 - val_acc: 0.9600\n",
"Epoch 407/10000\n",
"0s - loss: 0.2075 - acc: 0.9733 - val_loss: 0.2626 - val_acc: 0.9733\n",
"Epoch 408/10000\n",
"0s - loss: 0.2072 - acc: 0.9733 - val_loss: 0.2613 - val_acc: 0.9733\n",
"Epoch 409/10000\n",
"0s - loss: 0.2065 - acc: 0.9733 - val_loss: 0.2613 - val_acc: 0.9733\n",
"Epoch 410/10000\n",
"0s - loss: 0.2071 - acc: 0.9733 - val_loss: 0.2615 - val_acc: 0.9733\n",
"Epoch 411/10000\n",
"0s - loss: 0.2060 - acc: 0.9733 - val_loss: 0.2609 - val_acc: 0.9733\n",
"Epoch 412/10000\n",
"0s - loss: 0.2059 - acc: 0.9733 - val_loss: 0.2601 - val_acc: 0.9733\n",
"Epoch 413/10000\n",
"0s - loss: 0.2058 - acc: 0.9733 - val_loss: 0.2597 - val_acc: 0.9733\n",
"Epoch 414/10000\n",
"0s - loss: 0.2052 - acc: 0.9733 - val_loss: 0.2590 - val_acc: 0.9733\n",
"Epoch 415/10000\n",
"0s - loss: 0.2050 - acc: 0.9733 - val_loss: 0.2591 - val_acc: 0.9733\n",
"Epoch 416/10000\n",
"0s - loss: 0.2055 - acc: 0.9733 - val_loss: 0.2577 - val_acc: 0.9600\n",
"Epoch 417/10000\n",
"0s - loss: 0.2037 - acc: 0.9733 - val_loss: 0.2582 - val_acc: 0.9733\n",
"Epoch 418/10000\n",
"0s - loss: 0.2038 - acc: 0.9733 - val_loss: 0.2577 - val_acc: 0.9733\n",
"Epoch 419/10000\n",
"0s - loss: 0.2036 - acc: 0.9733 - val_loss: 0.2584 - val_acc: 0.9733\n"
]
}
],
"source": [
"early_stopping = EarlyStopping(patience=2)\n",
"history = model.fit(X, y, validation_split=0.5,\n",
" nb_epoch=10000, shuffle=True, batch_size=4,\n",
" verbose=2, callbacks=[early_stopping])"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Final Scores:\n",
"\tacc: 0.9733\n",
"\tloss: 0.2036\n",
"\tval_acc: 0.9733\n",
"\tval_loss: 0.2584\n",
"Num correct: 146 / 150\n"
]
}
],
"source": [
"print(\"Final Scores:\")\n",
"for k, v in sorted(history.history.items()):\n",
" print(\"\\t{}: {:0.4f}\".format(k, v[-1]))\n",
" \n",
"num_correct = sum(y.argmax(axis=1) == model.predict_classes(X, verbose=0))\n",
"print(\"Num correct: {} / {}\".format(num_correct, X.shape[0]))"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"collapsed": false
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAiIAAAFoCAYAAABqn5C9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQsTMzMwq8//4ZKLUgCXikQAAAABJRU5ErkJggg==\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x7f5dbbde0ba8>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plot_history(history);"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"collapsed": false
},
"outputs": [
{
"ename": "NameError",
"evalue": "name 'model' is not defined",
"output_type": "error",
"traceback": [
"\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
"\u001b[0;31mNameError\u001b[0m Traceback (most recent call last)",
"\u001b[0;32m<ipython-input-2-4c153b7b78f5>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlayers\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mW\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
"\u001b[0;31mNameError\u001b[0m: name 'model' is not defined"
]
}
],
"source": [
"model.layers[0].W"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.5.2+"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment