Datasets
Models
Downloads: 1
[a8e9b4]:
/
MultiscaleNet.ipynb
810 lines (809 with data), 31.9 kB
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"%matplotlib inline\n",
"import pandas as pd\n",
"import numpy as np\n",
"from cdtw import pydtw\n",
"import seaborn as sns\n",
"from tqdm import tqdm\n",
"import os\n",
"import json\n",
"import h5py"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Using TensorFlow backend.\n"
]
}
],
"source": [
"from keras import backend as K\n",
"from keras.regularizers import l2, activity_l2\n",
"\n",
"from keras import backend as K\n",
"from keras.engine.topology import Layer\n",
"from keras.optimizers import RMSprop, SGD, Adam\n",
"from keras.layers.core import Dense, Dropout, Activation, Flatten, Lambda, Merge\n",
"from keras.layers.recurrent import LSTM, GRU\n",
"from keras.models import Sequential, Model, load_model\n",
"from keras.layers import Input, Bidirectional, merge\n",
"from keras.layers.convolutional import Convolution1D, AtrousConvolution1D\n",
"from keras.layers.pooling import MaxPooling1D, AveragePooling1D, GlobalMaxPooling1D"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"def read_train(fname):\n",
" subjects = {}\n",
" with h5py.File(fname, \"r\") as data_file:\n",
" for subject, subject_data in data_file.items():\n",
" print(subject)\n",
" X = subject_data[\"data\"][:]\n",
" y = subject_data[\"labels\"][:]\n",
" subjects[subject] = (X, y)\n",
" return subjects\n",
"\n",
"def read_test(fname):\n",
" subjects = {}\n",
" with h5py.File(fname, \"r\") as data_file:\n",
" X = {}\n",
" for subject, subject_data in data_file.items():\n",
" X[subject] = {}\n",
" for chunk_id, chunk in data_file[subject].items():\n",
" X[subject][chunk_id] = chunk[:]\n",
"\n",
" return X\n",
"\n",
"def batch(ts, y, n=1):\n",
" l = len(ts)\n",
" for ndx in range(0, l-n, 1):\n",
" yield (ts[ndx:min(ndx + n, l)], y[ndx:min(ndx + n, l)])\n",
"\n",
"def label_batch(batch):\n",
" if all([i == 1 for i in batch[1]]):\n",
" return 1\n",
" elif all([i == 0 for i in batch[1]]):\n",
" return 0\n",
" elif all([i == 2 for i in batch[1]]):\n",
" return 2\n",
" return -1\n",
" \n",
"def get_data():\n",
" train = read_train(\"train.h5\")\n",
" test = read_test(\"test.h5\")\n",
" \n",
"\n",
" subject_datas = {}\n",
" for subject, data in tqdm(train.items()):\n",
" subject_ts = data[0].T\n",
" subject_y = data[1][0]\n",
" batches = [i for i in batch(subject_ts, subject_y, n=1125)]\n",
" batches = [(i[0], label_batch(i)) for i in batches]\n",
" batches = [i for i in batches if i[1] != -1]\n",
" batches = [i for i in batches if len(i[0]) == 1125]\n",
" subject_datas[subject] = batches\n",
" \n",
" X = []\n",
" y = []\n",
" for subj, subj_data in tqdm(subject_datas.items()):\n",
" X.extend([i[0] for i in subj_data])\n",
" y.extend([i[1] for i in subj_data])\n",
" return X, y, test"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"subject_0\n",
"subject_1\n",
"subject_2\n",
"subject_3\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 4/4 [01:14<00:00, 19.86s/it]\n",
"100%|██████████| 4/4 [00:00<00:00, 145.58it/s]\n"
]
}
],
"source": [
"X, y, test = get_data()"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {
"collapsed": false
},
"outputs": [],
"source": [
"X = np.array(X)\n",
"y = np.array(y)"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"def shuffle_in_unison_scary(a, b):\n",
" rng_state = np.random.get_state()\n",
" np.random.shuffle(a)\n",
" np.random.set_state(rng_state)\n",
" np.random.shuffle(b)"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"shuffle_in_unison_scary(X, y)"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {
"collapsed": false
},
"outputs": [],
"source": [
"validation_start = len(X) - 30\n",
"X_train = X[:validation_start]\n",
"y_train = y[:validation_start]\n",
"X_val = X[validation_start:]\n",
"y_val = y[validation_start:]"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"from scipy.signal import resample\n",
"\n",
"\n",
"def toarr(label):\n",
" arr = np.zeros(3)\n",
" arr[label] = 1\n",
" return arr\n",
"\n",
"def data_generator(X, Y, batch_size):\n",
" while True:\n",
" inds = np.random.choice(len(X), batch_size)\n",
" x = X[inds]\n",
" y = Y[inds]\n",
" y = np.vstack([toarr(i) for i in y])\n",
" x_256 = np.array([resample(i, 256) for i in x])\n",
" x_500 = np.array([resample(i, 500) for i in x])\n",
" x = np.array([i for i in x])\n",
" yield ([x_256, x_500, x], y)"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"def multiscale(chunk):\n",
" resampled_256 = resample(chunk, 256)\n",
" resampled_500 = resample(chunk, 500)\n",
" return [resampled_256, resampled_500, chunk]"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"def get_base_model(input_len, fsize):\n",
" '''Base network to be shared (eq. to feature extraction).\n",
" '''\n",
" with K.tf.device('/gpu:1'):\n",
" input_seq = Input(shape=(input_len, 24))\n",
" nb_filters = 150\n",
" convolved = Convolution1D(nb_filters, fsize, border_mode=\"same\", activation=\"tanh\")(input_seq)\n",
" processed = GlobalMaxPooling1D()(convolved)\n",
" compressed = Dense(150, activation=\"tanh\")(processed)\n",
" compressed = Dropout(0.3)(compressed)\n",
" compressed = Dense(150, activation=\"tanh\")(compressed)\n",
" model = Model(input=input_seq, output=compressed) \n",
" return model"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": false
},
"outputs": [],
"source": [
"with K.tf.device('/gpu:1'):\n",
" \n",
" input256_seq = Input(shape=(256, 24))\n",
" input500_seq = Input(shape=(500, 24))\n",
" input1125_seq = Input(shape=(1125, 24))\n",
" \n",
" base_network256 = get_base_model(256, 4)\n",
" base_network500 = get_base_model(500, 7)\n",
" base_network1125 = get_base_model(1125, 10)\n",
" \n",
" embedding_256 = base_network256(input256_seq)\n",
" embedding_500 = base_network500(input500_seq)\n",
" embedding_1125 = base_network256(input1125_seq)\n",
" \n",
" merged = merge([embedding_256, embedding_500, embedding_1125], mode=\"concat\")\n",
" out = Dense(3, activation='softmax')(merged)\n",
" \n",
" model = Model(input=[input256_seq, input500_seq, input1125_seq], output=out)\n",
" \n",
" #opt = SGD(lr=0.001, momentum=0.9, nesterov=True, clipvalue=0.0001)\n",
" opt = RMSprop(lr=0.005, clipvalue=10**6)\n",
" #opt = Adam(lr=0.001)\n",
" model.compile(loss=\"categorical_crossentropy\", optimizer=opt)"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"with K.tf.device('/gpu:2'):\n",
" model = load_model(\"convnet-multiscale-true-022unk\")"
]
},
{
"cell_type": "code",
"execution_count": 21,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/100000\n",
"100000/100000 [==============================] - 135s - loss: 0.1939 \n",
"Epoch 2/100000\n",
" 99500/100000 [============================>.] - ETA: 0s - loss: 0.1922"
]
},
{
"ename": "KeyboardInterrupt",
"evalue": "",
"output_type": "error",
"traceback": [
"\u001b[0;31m\u001b[0m",
"\u001b[0;31mKeyboardInterrupt\u001b[0mTraceback (most recent call last)",
"\u001b[0;32m<ipython-input-21-09ad3c297605>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m 6\u001b[0m \u001b[0;32mwith\u001b[0m \u001b[0mK\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdevice\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'/gpu:2'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 7\u001b[0m model.fit_generator(data_generator(X_train, y_train, batch_size=50), samples_per_epoch, nb_epoch, \n\u001b[0;32m----> 8\u001b[0;31m callbacks=[earlyStopping], verbose=1)#, nb_val_samples=20000,\n\u001b[0m\u001b[1;32m 9\u001b[0m \u001b[0;31m#validation_data=data_generator(X_val, y_val, batch_size=40))\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m/usr/local/lib/python2.7/dist-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mfit_generator\u001b[0;34m(self, generator, samples_per_epoch, nb_epoch, verbose, callbacks, validation_data, nb_val_samples, class_weight, max_q_size, nb_worker, pickle_safe)\u001b[0m\n\u001b[1;32m 1451\u001b[0m \u001b[0mbatch_logs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0ml\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mo\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 1452\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1453\u001b[0;31m \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_batch_end\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch_index\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_logs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 1454\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 1455\u001b[0m \u001b[0;31m# construct epoch logs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m/usr/local/lib/python2.7/dist-packages/keras/callbacks.pyc\u001b[0m in \u001b[0;36mon_batch_end\u001b[0;34m(self, batch, logs)\u001b[0m\n\u001b[1;32m 59\u001b[0m \u001b[0mt_before_callbacks\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 60\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mcallback\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcallbacks\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 61\u001b[0;31m \u001b[0mcallback\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_batch_end\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 62\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_delta_ts_batch_end\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mt_before_callbacks\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 63\u001b[0m \u001b[0mdelta_t_median\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmedian\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_delta_ts_batch_end\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m/usr/local/lib/python2.7/dist-packages/keras/callbacks.pyc\u001b[0m in \u001b[0;36mon_batch_end\u001b[0;34m(self, batch, logs)\u001b[0m\n\u001b[1;32m 187\u001b[0m \u001b[0;31m# will be handled by on_epoch_end\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 188\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mverbose\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mseen\u001b[0m \u001b[0;34m<\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mparams\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'nb_sample'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 189\u001b[0;31m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprogbar\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mupdate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mseen\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlog_values\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 190\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 191\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mon_epoch_end\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepoch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m{\u001b[0m\u001b[0;34m}\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m/usr/local/lib/python2.7/dist-packages/keras/utils/generic_utils.pyc\u001b[0m in \u001b[0;36mupdate\u001b[0;34m(self, current, values, force)\u001b[0m\n\u001b[1;32m 157\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 158\u001b[0m \u001b[0msys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstdout\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwrite\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minfo\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 159\u001b[0;31m \u001b[0msys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstdout\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mflush\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 160\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 161\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mcurrent\u001b[0m \u001b[0;34m>=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtarget\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m/usr/local/lib/python2.7/dist-packages/ipykernel/iostream.pyc\u001b[0m in \u001b[0;36mflush\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 267\u001b[0m \u001b[0mevt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mthreading\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mEvent\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 268\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_io_loop\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0madd_callback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mevt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 269\u001b[0;31m \u001b[0mevt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\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 270\u001b[0m \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 271\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_flush\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m/usr/lib/python2.7/threading.pyc\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m 618\u001b[0m \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 619\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__flag\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 620\u001b[0;31m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__cond\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 621\u001b[0m \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__flag\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 622\u001b[0m \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;32m/usr/lib/python2.7/threading.pyc\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m 337\u001b[0m \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;31m# restore state no matter what (e.g., KeyboardInterrupt)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 338\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 339\u001b[0;31m \u001b[0mwaiter\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macquire\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 340\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0m__debug__\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 341\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_note\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"%s.wait(): got it\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
"\u001b[0;31mKeyboardInterrupt\u001b[0m: "
]
}
],
"source": [
"from keras.callbacks import EarlyStopping\n",
"nb_epoch = 100000\n",
"earlyStopping = EarlyStopping(monitor='val_loss', patience=10, verbose=0, mode='auto')\n",
"samples_per_epoch = 100000\n",
"\n",
"with K.tf.device('/gpu:2'):\n",
" model.fit_generator(data_generator(X_train, y_train, batch_size=50), samples_per_epoch, nb_epoch, \n",
" callbacks=[earlyStopping], verbose=1)#, nb_val_samples=20000,\n",
" #validation_data=data_generator(X_val, y_val, batch_size=40))"
]
},
{
"cell_type": "code",
"execution_count": 26,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 70s - loss: 0.1640 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
" 24%|██▍ | 12/49 [00:00<00:00, 119.99it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 135.88it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 215.36it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 248.00it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 219.17it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 67s - loss: 0.1592 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"\r",
" 0%| | 0/49 [00:00<?, ?it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 126.00it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 190.92it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 161.84it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 190.56it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 60s - loss: 0.1557 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"\r",
" 0%| | 0/49 [00:00<?, ?it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 175.59it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 188.29it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 170.44it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 202.76it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 73s - loss: 0.1545 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
" 29%|██▊ | 14/49 [00:00<00:00, 139.19it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 190.23it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 235.47it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 221.90it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 171.74it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 72s - loss: 0.1556 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"\r",
" 0%| | 0/49 [00:00<?, ?it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 119.83it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 195.06it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 187.22it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 201.47it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 68s - loss: 0.1563 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
" 27%|██▋ | 13/49 [00:00<00:00, 119.17it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 142.53it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 203.55it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 192.02it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 193.69it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 72s - loss: 0.1588 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
" 31%|███ | 15/49 [00:00<00:00, 145.08it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 181.02it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 199.11it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 229.50it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 204.20it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 67s - loss: 0.1598 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
" 37%|███▋ | 18/49 [00:00<00:00, 174.26it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 130.09it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 208.28it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 189.43it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 181.52it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 53s - loss: 0.1634 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
" 41%|████ | 20/49 [00:00<00:00, 191.91it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 202.34it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 216.69it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 212.14it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 212.55it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/1\n",
"30000/30000 [==============================] - 70s - loss: 0.1491 "
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
" 29%|██▊ | 14/49 [00:00<00:00, 135.21it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 188.83it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 169.04it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 178.81it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 182.79it/s]\n"
]
}
],
"source": [
"# BLEND NNS\n",
"for blend_id in range(25, 35):\n",
" with K.tf.device('/gpu:2'):\n",
" model.fit_generator(data_generator(X_train, y_train, batch_size=100), samples_per_epoch=30000, nb_epoch=1, \n",
" callbacks=[earlyStopping], verbose=1)\n",
" \n",
" df = []\n",
" for subj in test:\n",
" for chunk in tqdm(test[subj]):\n",
" data = {}\n",
" data[\"subject_id\"] = int(subj.split(\"_\")[-1])\n",
" data[\"chunk_id\"] = int(chunk.split(\"_\")[-1])\n",
" arr = test[subj][chunk].T\n",
" preds = model.predict([np.array([i]) for i in multiscale(arr)])[0]\n",
" data[\"class_0_score\"] = preds[0]\n",
" data[\"class_1_score\"] = preds[1]\n",
" data[\"class_2_score\"] = preds[2]\n",
" for i in range(0, 1125):\n",
" data[\"tick\"] = i\n",
" df.append(data.copy())\n",
" df = pd.DataFrame(df)\n",
" df = df[[\"subject_id\", \"chunk_id\", \"tick\", \"class_0_score\",\n",
" \"class_1_score\",\"class_2_score\"]]\n",
" \n",
" df.to_csv('submit_blended_' + str(blend_id) + '.csv', index=False)"
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"model.save(\"convnet-multiscale-deep-021unk\")"
]
},
{
"cell_type": "code",
"execution_count": 19,
"metadata": {
"collapsed": false
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 49/49 [00:00<00:00, 165.60it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 185.89it/s]\n",
"100%|██████████| 49/49 [00:00<00:00, 189.62it/s]\n",
"100%|██████████| 89/89 [00:00<00:00, 175.60it/s]\n"
]
}
],
"source": [
"df = []\n",
"for subj in test:\n",
" for chunk in tqdm(test[subj]):\n",
" data = {}\n",
" data[\"subject_id\"] = int(subj.split(\"_\")[-1])\n",
" data[\"chunk_id\"] = int(chunk.split(\"_\")[-1])\n",
" arr = test[subj][chunk].T\n",
" preds = model.predict([np.array([i]) for i in multiscale(arr)])[0]\n",
" data[\"class_0_score\"] = preds[0]\n",
" data[\"class_1_score\"] = preds[1]\n",
" data[\"class_2_score\"] = preds[2]\n",
" for i in range(0, 1125):\n",
" data[\"tick\"] = i\n",
" df.append(data.copy())\n",
"df = pd.DataFrame(df)\n",
"df = df[[\"subject_id\", \"chunk_id\", \"tick\", \"class_0_score\",\n",
" \"class_1_score\",\"class_2_score\"]]"
]
},
{
"cell_type": "code",
"execution_count": 20,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"df.to_csv('submit_true_multiscale_016_large_batch.csv', index=False)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 2",
"language": "python",
"name": "python2"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 2
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython2",
"version": "2.7.6"
}
},
"nbformat": 4,
"nbformat_minor": 1
}