diff --git a/main/experiments/benchmark_vgg_end_to_end_new.py b/main/experiments/benchmark_vgg_end_to_end_new.py
new file mode 100644
index 0000000000000000000000000000000000000000..5e143e5a2659a1cebfed6c5e29a260c6e6980f30
--- /dev/null
+++ b/main/experiments/benchmark_vgg_end_to_end_new.py
@@ -0,0 +1,440 @@
+"""
+Benchmark VGG: Benchmarking deepstrom versus other architectures of the VGG network.
+
+Usage:
+ benchmark_vgg deepstrom [-r] [-a value] [-v size] [-e numepoch] [-s batchsize] [-D reprdim] [-m size] (-R|-L|-C|-E|-P|-S|-A|-T|-M) [-g gammavalue] [-c cvalue] [-n]
+
+Options:
+ --help -h Display help and exit.
+ -e numepoch --num-epoch=numepoch The number of epoch.
+ -s batchsize --batch-size=batchsize The number of example in each batch
+ -v size --validation-size size The size of the validation set [default: 10000]
+ -a value --seed value The seed value used for all randomization processed [default: 0]
+ -D reprdim --out-dim=reprdim The dimension of the final representation
+ -m size --nys-size size The number of example in the nystrom subsample.
+ -n --non-linear Tell Nystrom to use the non linear activation function on its output.
+ -r --real-nystrom Use the real w matrix
+ -g gammavalue --gamma gammavalue The value of gamma for rbf, chi or hyperbolic tangent kernel (deepstrom and deepfriedconvnet)
+ -c cvalue --intercept-constant cvalue The value of the intercept constant for the hyperbolic tangent kernel.
+ -R --rbf-kernel Says if the rbf kernel should be used for nystrom.
+ -L --linear-kernel Says if the linear kernel should be used for nystrom.
+ -C --chi-square-kernel Says if the basic additive chi square kernel should be used for nystrom.
+ -E --exp-chi-square-kernel Says if the exponential chi square kernel should be used for nystrom.
+ -P --chi-square-PD-kernel Says if the Positive definite version of the basic additive chi square kernel should be used for nystrom.
+ -S --sigmoid-kernel Says it the sigmoid kernel should be used for nystrom.
+ -A --laplacian-kernel Says if the laplacian kernel should be used for nystrom.
+ -T --stacked-kernel Says if the kernels laplacian, chi2 and rbf in a stacked setting should be used for nystrom.
+ -M --sumed-kernel Says if the kernels laplacian, chi2 and rbf in a summed setting should be used for nystrom.
+"""
+import sys
+import os
+import time as t
+import numpy as np
+import tensorflow as tf
+import docopt
+from keras import Model
+from keras.preprocessing.image import ImageDataGenerator
+import skluc.main.data.mldatasets as dataset
+from skluc.main.tensorflow_.kernel_approximation import nystrom_layer
+from skluc.main.tensorflow_.utils import batch_generator, classification_cifar
+from skluc.main.tensorflow_.kernel import tf_rbf_kernel, tf_linear_kernel, tf_chi_square_CPD, tf_chi_square_CPD_exp, \
+ tf_chi_square_PD, tf_sigmoid_kernel, tf_laplacian_kernel, tf_stack_of_kernels, tf_sum_of_kernels
+from skluc.main.utils import logger, log_memory_usage
+import keras
+from keras.models import Sequential, load_model
+from keras.layers import Activation
+from keras.layers import Conv2D, MaxPooling2D
+from keras.initializers import he_normal
+from keras.layers.normalization import BatchNormalization
+
+
+def VGG19(input_shape):
+ # with tf.variable_scope("block1_conv1"):
+ # weights = tf.get_variable("weights", (3, 3, 3, 64), initializer=tf.random_normal_initializer(stddev=0.1), trainable=trainable)
+ # biases = tf.get_variable("biases", (64), initializer=tf.constant_initializer(0.0), trainable=trainable)
+ # regularizer = tf.contrib.layers.l2_regularizer(scale=0.1)
+ # conv = tf.nn.conv2d(input_, weights, strides=[1, 1, 1, 1], padding='SAME', kernel_regularizer=regularizer)
+ # batch_norm = tf.nn.batch_normalization(conv, variance_epsilon=1e-3)
+ # relu = tf.nn.relu(conv + biases)
+ # tf.summary.histogram("act", relu)
+ # in order to reduce dimensionality, use bigger pooling size
+ # pool = max_pool(relu, pool_size=pool_size)
+ # with tf.variable_scope("conv_pool_2"):
+ # conv2 = conv_relu_pool(conv1, [5, 5, 6, 16], [16], pool_size=2, trainable=trainable)
+ weight_decay = 0.0001
+ # build model
+ model = Sequential()
+
+ # Block 1
+ model.add(Conv2D(64, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block1_conv1', input_shape=input_shape))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(64, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block1_conv2'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block1_pool'))
+
+ # Block 2
+ model.add(Conv2D(128, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block2_conv1'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(128, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block2_conv2'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block2_pool'))
+ #
+ # Block 3
+ model.add(Conv2D(256, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block3_conv1'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(256, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block3_conv2'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(256, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block3_conv3'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(256, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block3_conv4'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block3_pool'))
+ #
+ # Block 4
+ model.add(Conv2D(512, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block4_conv1'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(512, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block4_conv2'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(512, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block4_conv3'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(512, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block4_conv4'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block4_pool'))
+
+ # Block 5
+ model.add(Conv2D(512, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block5_conv1'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(512, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block5_conv2'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(512, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block5_conv3'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(Conv2D(512, (3, 3), padding='same', kernel_regularizer=keras.regularizers.l2(weight_decay), kernel_initializer=he_normal(), name='block5_conv4'))
+ model.add(BatchNormalization())
+ model.add(Activation('relu'))
+ model.add(MaxPooling2D((2, 2), strides=(2, 2), name='block5_pool'))
+
+ return model
+
+
+def VGG19_preload():
+ logger.debug("filename: {}".format(os.path.abspath(__file__)))
+ model = load_model(os.path.join(os.path.dirname(os.path.abspath(__file__)), "1522967518.1916964_vgg19_cifar10.h5"))
+ vgg_conv_model = Model(inputs=model.input,
+ outputs=model.get_layer('block5_pool').output)
+ return vgg_conv_model
+
+
+def get_gamma_value(arguments, dat, chi2=False):
+ if arguments["--gamma"] is None:
+ logger.debug("Gamma arguments is None. Need to compute it.")
+ if chi2:
+ gamma_value = 1./compute_euristic_sigma_chi2(dat.train.data)
+
+ else:
+ gamma_value = 1./compute_euristic_sigma(dat.train.data)
+ else:
+ gamma_value = eval(arguments["--gamma"])
+
+ logger.debug("Gamma value is {}".format(gamma_value))
+ return gamma_value
+
+def init_kernel():
+ kernel_dict = {}
+ GAMMA = None
+ if RBF_KERNEL:
+ KERNEL = tf_rbf_kernel
+ KERNEL_NAME = "rbf"
+ GAMMA = get_gamma_value(arguments, data)
+ kernel_dict = {"gamma": GAMMA}
+ elif LINEAR_KERNEL:
+ KERNEL = tf_linear_kernel
+ KERNEL_NAME = "linear"
+ elif CHI2_KERNEL:
+ KERNEL = tf_chi_square_CPD
+ KERNEL_NAME = "chi2_cpd"
+ elif CHI2_EXP_KERNEL:
+ KERNEL = tf_chi_square_CPD_exp
+ KERNEL_NAME = "chi2_exp_cpd"
+ GAMMA = get_gamma_value(arguments, data, chi2=True)
+ kernel_dict = {"gamma": GAMMA}
+ elif CHI2_PD_KERNEL:
+ KERNEL = tf_chi_square_PD
+ KERNEL_NAME = "chi2_pd"
+ elif SIGMOID_KERNEL:
+ KERNEL = tf_sigmoid_kernel
+ KERNEL_NAME = "sigmoid"
+ GAMMA = get_gamma_value(arguments, data)
+ CONST = float(arguments["--intercept-constant"])
+ kernel_dict = {"gamma": GAMMA, "constant": CONST}
+ elif LAPLACIAN_KERNEL:
+ KERNEL = tf_laplacian_kernel
+ KERNEL_NAME = "laplacian"
+ GAMMA = get_gamma_value(arguments, data)
+ kernel_dict = {"gamma": np.sqrt(GAMMA)}
+ elif STACKED_KERNEL:
+ GAMMA = get_gamma_value(arguments, data)
+
+
+ def KERNEL(X, Y):
+ return tf_stack_of_kernels(X, Y, [tf_rbf_kernel for _ in GAMMA],
+ [{"gamma": g_value} for g_value in GAMMA])
+
+
+ KERNEL_NAME = "stacked"
+
+ elif SUMED_KERNEL:
+ GAMMA = get_gamma_value(arguments, data)
+
+
+ def KERNEL(X, Y):
+ return tf_sum_of_kernels(X, Y, [tf_rbf_kernel for _ in GAMMA],
+ [{"gamma": g_value} for g_value in GAMMA])
+
+
+ KERNEL_NAME = "summed"
+ else:
+ raise Exception("No kernel function specified for deepstrom")
+
+ return KERNEL_NAME, KERNEL, kernel_dict, GAMMA
+
+
+if __name__ == '__main__':
+
+ arguments = docopt.docopt(__doc__)
+ NUM_EPOCH = int(arguments["--num-epoch"])
+ BATCH_SIZE = int(arguments["--batch-size"])
+ SEED_TRAIN_VALIDATION = 0
+ SEED = int(arguments["--seed"])
+ OUT_DIM = int(arguments["--out-dim"]) if arguments["--out-dim"] is not None else None
+ VALIDATION_SIZE = int(arguments["--validation-size"])
+ NYS_SUBSAMPLE_SIZE = int(arguments["--nys-size"])
+ if OUT_DIM is None:
+ OUT_DIM = NYS_SUBSAMPLE_SIZE
+ KERNEL_NAME = None
+ GAMMA = None
+ CONST = None
+ REAL_NYSTROM = arguments["--real-nystrom"]
+
+ NON_LINEAR = tf.nn.relu if arguments["--non-linear"] else None
+
+ RBF_KERNEL = arguments["--rbf-kernel"]
+ LINEAR_KERNEL = arguments["--linear-kernel"]
+ CHI2_KERNEL = arguments["--chi-square-kernel"]
+ CHI2_EXP_KERNEL = arguments["--exp-chi-square-kernel"]
+ CHI2_PD_KERNEL = arguments["--chi-square-PD-kernel"]
+ SIGMOID_KERNEL = arguments["--sigmoid-kernel"]
+ LAPLACIAN_KERNEL = arguments["--laplacian-kernel"]
+ STACKED_KERNEL = arguments["--stacked-kernel"]
+ SUMED_KERNEL = arguments["--sumed-kernel"]
+
+ data = dataset.Cifar10Dataset(validation_size=VALIDATION_SIZE, seed=SEED_TRAIN_VALIDATION)
+ data.load()
+ data.normalize()
+ data.data_astype(np.float32)
+ data.labels_astype(np.float32)
+ data.to_image()
+ data.to_one_hot()
+
+ logger.debug("Start benchmark with parameters: {}".format(" ".join(sys.argv[1:])))
+ logger.debug("Using dataset {} with validation size {} and seed for spliting set {}.".format(data.s_name, data.validation_size, data.seed))
+ logger.debug("Shape of train set data: {}; shape of train set labels: {}".format(data.train[0].shape, data.train[1].shape))
+ logger.debug("Shape of validation set data: {}; shape of validation set labels: {}".format(data.validation[0].shape, data.validation[1].shape))
+ logger.debug("Shape of test set data: {}; shape of test set labels: {}".format(data.test[0].shape, data.test[1].shape))
+ logger.debug("Sample of label: {}".format(data.train[1][0]))
+
+ NYS_SUBSAMPLE_SIZE = int(arguments["--nys-size"])
+ if OUT_DIM is None:
+ OUT_DIM = NYS_SUBSAMPLE_SIZE
+
+ KERNEL_NAME, KERNEL, kernel_dict, GAMMA = init_kernel()
+
+ input_dim, output_dim = data.train[0].shape[1:], data.train[1].shape[1]
+ with tf.Graph().as_default():
+ np.random.seed(SEED)
+ nys_subsample_index = np.random.permutation(data.train[0].shape[0])
+ nys_subsample = data.train[0][nys_subsample_index[:NYS_SUBSAMPLE_SIZE]]
+
+ nys_subsample_placeholder = tf.Variable(nys_subsample, dtype=tf.float32, name="nys_subsample", trainable=False)
+
+ x = tf.placeholder(tf.float32, shape=[None, *input_dim], name="x")
+ y = tf.placeholder(tf.float32, shape=[None, output_dim], name="label")
+ # nys_subsample_placeholder = tf.placeholder(tf.float32, shape=[NYS_SUBSAMPLE_SIZE, *input_dim], name="nys_subsample")
+
+ # vgg_conv_model = VGG19_preload()
+ with tf.variable_scope("Convolution") as scope_convolution:
+ vgg_conv_model = VGG19(input_dim)
+ vgg_conv_model.trainable=False
+ conv_x = vgg_conv_model(x)
+ tf.summary.histogram("convolution_x", conv_x)
+ vgg_conv_model_subsample = keras.Model(inputs=vgg_conv_model.inputs,
+ outputs=vgg_conv_model.outputs)
+ vgg_conv_model_subsample.trainable = False
+ conv_nys_subsample = vgg_conv_model_subsample(nys_subsample_placeholder)
+
+ logger.debug("Selecting deepstrom layer function with "
+ "subsample size = {}, "
+ "output_dim = {}, "
+ "{} activation function "
+ "and kernel = {}"
+ .format(NYS_SUBSAMPLE_SIZE,
+ OUT_DIM,
+ "with" if NON_LINEAR else "without",
+ KERNEL_NAME))
+ if OUT_DIM is not None and OUT_DIM > NYS_SUBSAMPLE_SIZE:
+ logger.debug("Output dim is greater than deepstrom subsample size. Aborting.")
+ # todo change this because it is copy-pasted (use function instead)
+
+ global_acc_val = None
+ global_acc_test = None
+ training_time = None
+ printed_r_list = [str(global_acc_val),
+ str(global_acc_test),
+ str(training_time),
+ str(NUM_EPOCH),
+ str(BATCH_SIZE),
+ str(OUT_DIM),
+ str(KERNEL_NAME),
+ str(GAMMA),
+ str(CONST),
+ str(NYS_SUBSAMPLE_SIZE),
+ str(VALIDATION_SIZE),
+ str(SEED),
+ str(NON_LINEAR),
+ ]
+ print(",".join(printed_r_list))
+ exit()
+ w_matrix = None
+ if REAL_NYSTROM:
+ init_dim = np.prod([s.value for s in conv_x.shape[1:] if s.value is not None])
+ h_conv_nystrom_subsample_flat = tf.reshape(conv_nys_subsample, [conv_nys_subsample.shape[0], init_dim])
+
+ K_matrix = KERNEL(h_conv_nystrom_subsample_flat, h_conv_nystrom_subsample_flat, **kernel_dict)
+ S, U, V = tf.svd(K_matrix)
+ invert_root_K = tf.matmul(tf.matmul(U, tf.sqrt(tf.diag(S))), tf.transpose(V))
+ w_matrix = invert_root_K
+
+ input_classif = fct_deepstrom(conv_x, OUT_DIM, conv_nys_subsample, KERNEL, kernel_dict, w_matrix=w_matrix, non_linearity=NON_LINEAR)
+
+ classif, keep_prob = classification_cifar(input_classif, output_dim)
+
+ # calcul de la loss
+ with tf.name_scope("xent"):
+ cross_entropy = tf.reduce_mean(
+ tf.nn.softmax_cross_entropy_with_logits(labels=y, logits=classif, name="xentropy"),
+ name="xentropy_mean")
+ tf.summary.scalar('loss-xent', cross_entropy)
+
+ # todo learning rate as hyperparameter
+ # calcul du gradient
+ with tf.name_scope("train"):
+ global_step = tf.Variable(0, name="global_step", trainable=False)
+ train_optimizer = tf.train.AdamOptimizer(learning_rate=1e-4).minimize(cross_entropy,
+ global_step=global_step)
+
+ # calcul de l'accuracy
+ with tf.name_scope("accuracy"):
+ predictions = tf.argmax(classif, 1)
+ correct_prediction = tf.equal(predictions, tf.argmax(y, 1))
+ accuracy_op = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
+ tf.summary.scalar("accuracy", accuracy_op)
+
+ merged_summary = tf.summary.merge_all()
+
+ init = tf.global_variables_initializer()
+ # Create a session for running Ops on the Graph.
+ # Instantiate a SummaryWriter to output summaries and the Graph.
+ # summary_writer = tf.summary.FileWriter("debug_benchmark_vgg")
+ # Initialize all Variable objects
+ # actual learning
+ saver = tf.train.Saver()
+
+ with tf.Session() as sess:
+ logger.debug("trainable variables are: {}".format(tf.trainable_variables()))
+ # summary_writer.add_graph(sess.graph)
+ # Initialize all Variable objects
+ datagen = ImageDataGenerator(horizontal_flip=True,
+ width_shift_range=0.125,
+ height_shift_range=0.125,
+ fill_mode='constant',
+ cval=0.)
+ datagen.fit(data.train[0])
+ sess.run(init)
+ # actual learning
+ # feed_dict_val = {x: data.validation[0], y: data.validation[1], keep_prob: 1.0}
+ global_start = t.time()
+ feed_dict = {nys_subsample_placeholder: nys_subsample}
+ feed_dict_val = {nys_subsample_placeholder: nys_subsample}
+ feed_dict_test = {nys_subsample_placeholder: nys_subsample}
+ start_time_int = int(t.time())
+ for i in range(NUM_EPOCH):
+ saver.save(sess, os.path.abspath('end_to_end_model'), global_step=start_time_int)
+ start = t.time()
+ # for X_batch, Y_batch in batch_generator(data.train[0], data.train[1], BATCH_SIZE, True):
+ batchgen = datagen.flow(data.train[0], data.train[1], BATCH_SIZE, shuffle=False)
+ j = 0
+ log_memory_usage()
+ while j < len(batchgen):
+ X_batch, Y_batch = next(batchgen)
+ # batch_generator(data.train[0], data.train[1], BATCH_SIZE, True):
+ # X_batch = tf.map_fn(lambda img: datagen.random_transform(img), X_batch)
+ feed_dict.update({x: X_batch, y: Y_batch, keep_prob: 0.5})
+ _, loss, acc = sess.run([train_optimizer, cross_entropy, accuracy_op], feed_dict=feed_dict)
+ if j % 100 == 0:
+ # summary_str = sess.run(merged_summary, feed_dict=feed_dict)
+ # summary_writer.add_summary(summary_str, j)
+ logger.debug("epoch: {}/{}; batch: {}/{}; loss: {}; acc: {}".format(i, NUM_EPOCH,
+ j, int(data.train[0].shape[0]/BATCH_SIZE),
+ loss, acc))
+ j += 1
+
+ training_time = t.time() - global_start
+ accuracies_val = []
+ i = 0
+ for X_batch, Y_batch in batch_generator(data.validation[0], data.validation[1], 1000, False):
+ feed_dict_val.update({x: X_batch, y: Y_batch, keep_prob: 1.0})
+ accuracy = sess.run([accuracy_op], feed_dict=feed_dict_val)
+ accuracies_val.append(accuracy[0])
+ i += 1
+
+ accuracies_test = []
+ i = 0
+ for X_batch, Y_batch in batch_generator(data.test[0], data.test[1], 1000, False):
+ feed_dict_test.update({x: X_batch, y: Y_batch, keep_prob: 1.0})
+ accuracy = sess.run([accuracy_op], feed_dict=feed_dict_test)
+ accuracies_test.append(accuracy[0])
+ i += 1
+
+ global_acc_val = sum(accuracies_val) / i
+ global_acc_test = sum(accuracies_test) / i
+ printed_r_list = [str(global_acc_val),
+ str(global_acc_test),
+ str(training_time),
+ str(NUM_EPOCH),
+ str(BATCH_SIZE),
+ str(OUT_DIM),
+ str(KERNEL_NAME),
+ str(GAMMA),
+ str(CONST),
+ str(NYS_SUBSAMPLE_SIZE),
+ str(VALIDATION_SIZE),
+ str(SEED),
+ str(NON_LINEAR),
+ ]
+ print(",".join(printed_r_list))
+
+
diff --git a/main/experiments/deepstrom_classif_end_to_end_mnist.ipynb b/main/experiments/deepstrom_classif_end_to_end_mnist.ipynb
new file mode 100644
index 0000000000000000000000000000000000000000..01c10b25b30536ce51dd9b5d827ca16f8237e65a
--- /dev/null
+++ b/main/experiments/deepstrom_classif_end_to_end_mnist.ipynb
@@ -0,0 +1,34 @@
+{
+ "cells": [
+ {
+ "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": 0
+}
\ No newline at end of file
diff --git a/main/experiments/graph_drawing/paper/cifar/cifar_few_data_more_subsample.zip b/main/experiments/graph_drawing/paper/cifar/cifar_few_data_more_subsample.zip
deleted file mode 100644
index 010806ef26483a5ab635d653441d230fd96318bc..0000000000000000000000000000000000000000
Binary files a/main/experiments/graph_drawing/paper/cifar/cifar_few_data_more_subsample.zip and /dev/null differ
diff --git a/main/experiments/graph_drawing/paper/cifar/cifar_sota.zip b/main/experiments/graph_drawing/paper/cifar/cifar_sota.zip
deleted file mode 100644
index a9b058c904d7cd6aaa21cae5a4272d4d79f95bb9..0000000000000000000000000000000000000000
Binary files a/main/experiments/graph_drawing/paper/cifar/cifar_sota.zip and /dev/null differ
diff --git a/main/experiments/graph_drawing/paper/cifar/multikernel/by_sigma_value.zip b/main/experiments/graph_drawing/paper/cifar/multikernel/by_sigma_value.zip
deleted file mode 100644
index 744cde3f3c373edee67ec0d4715d122ac8189612..0000000000000000000000000000000000000000
Binary files a/main/experiments/graph_drawing/paper/cifar/multikernel/by_sigma_value.zip and /dev/null differ
diff --git a/main/experiments/graph_drawing/paper/cifar/plt_2D.zip b/main/experiments/graph_drawing/paper/cifar/plt_2D.zip
deleted file mode 100644
index 7cc268b56104a4ff05d0dfee07346ec0217d042e..0000000000000000000000000000000000000000
Binary files a/main/experiments/graph_drawing/paper/cifar/plt_2D.zip and /dev/null differ
diff --git a/main/experiments/graph_drawing/paper/mnist/mnist_few_data_more_subsample.zip b/main/experiments/graph_drawing/paper/mnist/mnist_few_data_more_subsample.zip
deleted file mode 100644
index beb2bd587f37e5580fb8edba6a108e5ad3d9c563..0000000000000000000000000000000000000000
Binary files a/main/experiments/graph_drawing/paper/mnist/mnist_few_data_more_subsample.zip and /dev/null differ
diff --git a/main/experiments/graph_drawing/paper/mnist/mnist_sota.zip b/main/experiments/graph_drawing/paper/mnist/mnist_sota.zip
deleted file mode 100644
index c5177ae60b2a5909796ad2804e99069803b37a89..0000000000000000000000000000000000000000
Binary files a/main/experiments/graph_drawing/paper/mnist/mnist_sota.zip and /dev/null differ
diff --git a/main/experiments/graph_drawing/paper/__init__.py b/main/experiments/graph_drawing/till_october_2018/__init__.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/__init__.py
rename to main/experiments/graph_drawing/till_october_2018/__init__.py
diff --git a/main/experiments/graph_drawing/paper/cifar100/__init__.py b/main/experiments/graph_drawing/till_october_2018/cifar100_bloc3_test/__init__.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/cifar100/__init__.py
rename to main/experiments/graph_drawing/till_october_2018/cifar100_bloc3_test/__init__.py
diff --git a/main/experiments/graph_drawing/cifar100_bloc3_test/cifar_100_bloc3_test.py b/main/experiments/graph_drawing/till_october_2018/cifar100_bloc3_test/cifar_100_bloc3_test.py
similarity index 100%
rename from main/experiments/graph_drawing/cifar100_bloc3_test/cifar_100_bloc3_test.py
rename to main/experiments/graph_drawing/till_october_2018/cifar100_bloc3_test/cifar_100_bloc3_test.py
diff --git a/main/experiments/graph_drawing/paper_iclr/__init__.py b/main/experiments/graph_drawing/till_october_2018/cifar100_cifar10_svhn_2_layers_non_lin/__init__.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/__init__.py
rename to main/experiments/graph_drawing/till_october_2018/cifar100_cifar10_svhn_2_layers_non_lin/__init__.py
diff --git a/main/experiments/graph_drawing/cifar100_cifar10_svhn_2_layers_non_lin/cifar_100_all_data.py b/main/experiments/graph_drawing/till_october_2018/cifar100_cifar10_svhn_2_layers_non_lin/cifar_100_all_data.py
similarity index 100%
rename from main/experiments/graph_drawing/cifar100_cifar10_svhn_2_layers_non_lin/cifar_100_all_data.py
rename to main/experiments/graph_drawing/till_october_2018/cifar100_cifar10_svhn_2_layers_non_lin/cifar_100_all_data.py
diff --git a/main/experiments/graph_drawing/draw_graphes.py b/main/experiments/graph_drawing/till_october_2018/draw_graphes.py
similarity index 100%
rename from main/experiments/graph_drawing/draw_graphes.py
rename to main/experiments/graph_drawing/till_october_2018/draw_graphes.py
diff --git a/main/experiments/graph_drawing/draw_graphes_vgg_cifar.py b/main/experiments/graph_drawing/till_october_2018/draw_graphes_vgg_cifar.py
similarity index 100%
rename from main/experiments/graph_drawing/draw_graphes_vgg_cifar.py
rename to main/experiments/graph_drawing/till_october_2018/draw_graphes_vgg_cifar.py
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_normal/__init__.py b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_big_subsample/__init__.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_normal/__init__.py
rename to main/experiments/graph_drawing/till_october_2018/few_shot_tests_big_subsample/__init__.py
diff --git a/main/experiments/graph_drawing/few_shot_tests_big_subsample/graph_drawing_few_shot_tests_big_subsample.ipynb b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_big_subsample/graph_drawing_few_shot_tests_big_subsample.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/few_shot_tests_big_subsample/graph_drawing_few_shot_tests_big_subsample.ipynb
rename to main/experiments/graph_drawing/till_october_2018/few_shot_tests_big_subsample/graph_drawing_few_shot_tests_big_subsample.ipynb
diff --git a/main/experiments/graph_drawing/few_shot_tests_big_subsample/graph_drawing_few_shot_tests_big_subsample.py b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_big_subsample/graph_drawing_few_shot_tests_big_subsample.py
similarity index 100%
rename from main/experiments/graph_drawing/few_shot_tests_big_subsample/graph_drawing_few_shot_tests_big_subsample.py
rename to main/experiments/graph_drawing/till_october_2018/few_shot_tests_big_subsample/graph_drawing_few_shot_tests_big_subsample.py
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_normal/cifar10/__init__.py b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_knn_nobacktrain/__init__.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_normal/cifar10/__init__.py
rename to main/experiments/graph_drawing/till_october_2018/few_shot_tests_knn_nobacktrain/__init__.py
diff --git a/main/experiments/graph_drawing/few_shot_tests_knn_nobacktrain/graph_drawing_few_shot_tests_knn_nobacktrain.ipynb b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_knn_nobacktrain/graph_drawing_few_shot_tests_knn_nobacktrain.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/few_shot_tests_knn_nobacktrain/graph_drawing_few_shot_tests_knn_nobacktrain.ipynb
rename to main/experiments/graph_drawing/till_october_2018/few_shot_tests_knn_nobacktrain/graph_drawing_few_shot_tests_knn_nobacktrain.ipynb
diff --git a/main/experiments/graph_drawing/few_shot_tests_knn_nobacktrain/graph_drawing_few_shot_tests_knn_nobacktrain.py b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_knn_nobacktrain/graph_drawing_few_shot_tests_knn_nobacktrain.py
similarity index 100%
rename from main/experiments/graph_drawing/few_shot_tests_knn_nobacktrain/graph_drawing_few_shot_tests_knn_nobacktrain.py
rename to main/experiments/graph_drawing/till_october_2018/few_shot_tests_knn_nobacktrain/graph_drawing_few_shot_tests_knn_nobacktrain.py
diff --git a/main/experiments/graph_drawing/till_october_2018/few_shot_tests_proto_withW/__init__.py b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_proto_withW/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/few_shot_tests_proto_withW/graph_drawing_few_shot_tests_proto_withW.ipynb b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_proto_withW/graph_drawing_few_shot_tests_proto_withW.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/few_shot_tests_proto_withW/graph_drawing_few_shot_tests_proto_withW.ipynb
rename to main/experiments/graph_drawing/till_october_2018/few_shot_tests_proto_withW/graph_drawing_few_shot_tests_proto_withW.ipynb
diff --git a/main/experiments/graph_drawing/few_shot_tests_proto_withW/graph_drawing_few_shot_tests_proto_withW.py b/main/experiments/graph_drawing/till_october_2018/few_shot_tests_proto_withW/graph_drawing_few_shot_tests_proto_withW.py
similarity index 100%
rename from main/experiments/graph_drawing/few_shot_tests_proto_withW/graph_drawing_few_shot_tests_proto_withW.py
rename to main/experiments/graph_drawing/till_october_2018/few_shot_tests_proto_withW/graph_drawing_few_shot_tests_proto_withW.py
diff --git a/main/experiments/graph_drawing/till_october_2018/multiview_cifar100/__init__.py b/main/experiments/graph_drawing/till_october_2018/multiview_cifar100/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/multiview_cifar100/multiview_cifar100.py b/main/experiments/graph_drawing/till_october_2018/multiview_cifar100/multiview_cifar100.py
similarity index 100%
rename from main/experiments/graph_drawing/multiview_cifar100/multiview_cifar100.py
rename to main/experiments/graph_drawing/till_october_2018/multiview_cifar100/multiview_cifar100.py
diff --git a/main/experiments/graph_drawing/multiview_cifar100_lin_no_w/graph_drawing_multiview_cifar100_no_w.ipynb b/main/experiments/graph_drawing/till_october_2018/multiview_cifar100_lin_no_w/graph_drawing_multiview_cifar100_no_w.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/multiview_cifar100_lin_no_w/graph_drawing_multiview_cifar100_no_w.ipynb
rename to main/experiments/graph_drawing/till_october_2018/multiview_cifar100_lin_no_w/graph_drawing_multiview_cifar100_no_w.ipynb
diff --git a/main/experiments/graph_drawing/till_october_2018/multiview_svhn/__init__.py b/main/experiments/graph_drawing/till_october_2018/multiview_svhn/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/multiview_svhn/multiview_svhn.py b/main/experiments/graph_drawing/till_october_2018/multiview_svhn/multiview_svhn.py
similarity index 100%
rename from main/experiments/graph_drawing/multiview_svhn/multiview_svhn.py
rename to main/experiments/graph_drawing/till_october_2018/multiview_svhn/multiview_svhn.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/till_october_2018/paper/cifar100/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper/cifar100/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper/cifar100/cifar_100_all_data.py b/main/experiments/graph_drawing/till_october_2018/paper/cifar100/cifar_100_all_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/cifar100/cifar_100_all_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper/cifar100/cifar_100_all_data.py
diff --git a/main/experiments/graph_drawing/paper/cifar100/cifar_100_all_data_time.py b/main/experiments/graph_drawing/till_october_2018/paper/cifar100/cifar_100_all_data_time.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/cifar100/cifar_100_all_data_time.py
rename to main/experiments/graph_drawing/till_october_2018/paper/cifar100/cifar_100_all_data_time.py
diff --git a/main/experiments/graph_drawing/paper/cifar100/cifar_100_few_data.py b/main/experiments/graph_drawing/till_october_2018/paper/cifar100/cifar_100_few_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/cifar100/cifar_100_few_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper/cifar100/cifar_100_few_data.py
diff --git a/main/experiments/graph_drawing/paper/tcnn_moviereview_pre_trained_expe.py b/main/experiments/graph_drawing/till_october_2018/paper/tcnn_moviereview_pre_trained_expe.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/tcnn_moviereview_pre_trained_expe.py
rename to main/experiments/graph_drawing/till_october_2018/paper/tcnn_moviereview_pre_trained_expe.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper/very_big_all_datasets_few_data/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper/very_big_all_datasets_few_data/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper/very_big_all_datasets_few_data/all_dataset_few_data.py b/main/experiments/graph_drawing/till_october_2018/paper/very_big_all_datasets_few_data/all_dataset_few_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/very_big_all_datasets_few_data/all_dataset_few_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper/very_big_all_datasets_few_data/all_dataset_few_data.py
diff --git a/main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_expe_bad_vgg_124_conv.py b/main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_expe_bad_vgg_124_conv.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_expe_bad_vgg_124_conv.py
rename to main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_expe_bad_vgg_124_conv.py
diff --git a/main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_expe_bad_vgg_3_conv.py b/main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_expe_bad_vgg_3_conv.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_expe_bad_vgg_3_conv.py
rename to main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_expe_bad_vgg_3_conv.py
diff --git a/main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_expe_few_data.py b/main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_expe_few_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_expe_few_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_expe_few_data.py
diff --git a/main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_stacked_summed.py b/main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_stacked_summed.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_stacked_summed.py
rename to main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_stacked_summed.py
diff --git a/main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_stacked_summed_sigma_range.py b/main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_stacked_summed_sigma_range.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/vgg_cifar_deepstrom_pre_trained_stacked_summed_sigma_range.py
rename to main/experiments/graph_drawing/till_october_2018/paper/vgg_cifar_deepstrom_pre_trained_stacked_summed_sigma_range.py
diff --git a/main/experiments/graph_drawing/paper/vgg_svhn_deepstrom_few_data.py b/main/experiments/graph_drawing/till_october_2018/paper/vgg_svhn_deepstrom_few_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/vgg_svhn_deepstrom_few_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper/vgg_svhn_deepstrom_few_data.py
diff --git a/main/experiments/graph_drawing/paper/vgg_svhn_deepstrom_std_classif.py b/main/experiments/graph_drawing/till_october_2018/paper/vgg_svhn_deepstrom_std_classif.py
similarity index 100%
rename from main/experiments/graph_drawing/paper/vgg_svhn_deepstrom_std_classif.py
rename to main/experiments/graph_drawing/till_october_2018/paper/vgg_svhn_deepstrom_std_classif.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_multiview/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_multiview/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_multiview/multiview_cifar100.ipynb b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_multiview/multiview_cifar100.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_multiview/multiview_cifar100.ipynb
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_multiview/multiview_cifar100.ipynb
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_multiview/multiview_cifar100.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_multiview/multiview_cifar100.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_multiview/multiview_cifar100.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_multiview/multiview_cifar100.py
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_multiview/resultsmax.ods b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_multiview/resultsmax.ods
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_multiview/resultsmax.ods
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_multiview/resultsmax.ods
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/cifar10/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/cifar10/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_normal/cifar10/vgg_cifar_deepstrom_classif_normal.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/cifar10/vgg_cifar_deepstrom_classif_normal.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_normal/cifar10/vgg_cifar_deepstrom_classif_normal.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/cifar10/vgg_cifar_deepstrom_classif_normal.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/cifar100/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/cifar100/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_normal/cifar100/vgg_cifar100_deepstrom_classif_normal.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/cifar100/vgg_cifar100_deepstrom_classif_normal.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_normal/cifar100/vgg_cifar100_deepstrom_classif_normal.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/cifar100/vgg_cifar100_deepstrom_classif_normal.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/mnist/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/mnist/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_normal/mnist/vgg_mnist_deepstrom_classif_normal.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/mnist/vgg_mnist_deepstrom_classif_normal.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_normal/mnist/vgg_mnist_deepstrom_classif_normal.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/mnist/vgg_mnist_deepstrom_classif_normal.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/svhn/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/svhn/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_normal/svhn/vgg_svhn_deepstrom_classif_normal.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/svhn/vgg_svhn_deepstrom_classif_normal.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_normal/svhn/vgg_svhn_deepstrom_classif_normal.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_normal/svhn/vgg_svhn_deepstrom_classif_normal.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar10/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar10/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/cifar10/vgg_cifar10_smll_dataset.ipynb b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar10/vgg_cifar10_smll_dataset.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/cifar10/vgg_cifar10_smll_dataset.ipynb
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar10/vgg_cifar10_smll_dataset.ipynb
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/cifar10/vgg_cifar_deepstrom_few_data.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar10/vgg_cifar_deepstrom_few_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/cifar10/vgg_cifar_deepstrom_few_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar10/vgg_cifar_deepstrom_few_data.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar100/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar100/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/cifar100/vgg_cifar100.ipynb b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar100/vgg_cifar100.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/cifar100/vgg_cifar100.ipynb
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar100/vgg_cifar100.ipynb
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/cifar100/vgg_deepstrom_cifar100_few_data.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar100/vgg_deepstrom_cifar100_few_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/cifar100/vgg_deepstrom_cifar100_few_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/cifar100/vgg_deepstrom_cifar100_few_data.py
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/mnist/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/mnist/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/mnist/lecun_mnist.ipynb b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/mnist/lecun_mnist.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/mnist/lecun_mnist.ipynb
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/mnist/lecun_mnist.ipynb
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/mnist/vgg_mnist_deepstrom_few_data.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/mnist/vgg_mnist_deepstrom_few_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/mnist/vgg_mnist_deepstrom_few_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/mnist/vgg_mnist_deepstrom_few_data.py
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/resultsmax.ods b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/resultsmax.ods
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/resultsmax.ods
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/resultsmax.ods
diff --git a/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/svhn/__init__.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/svhn/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/svhn/vgg_svhn.ipynb b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/svhn/vgg_svhn.ipynb
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/svhn/vgg_svhn.ipynb
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/svhn/vgg_svhn.ipynb
diff --git a/main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/svhn/vgg_svhn_deepstrom_few_data.py b/main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/svhn/vgg_svhn_deepstrom_few_data.py
similarity index 100%
rename from main/experiments/graph_drawing/paper_iclr/accuracy_small_dataset/svhn/vgg_svhn_deepstrom_few_data.py
rename to main/experiments/graph_drawing/till_october_2018/paper_iclr/accuracy_small_dataset/svhn/vgg_svhn_deepstrom_few_data.py
diff --git a/main/experiments/graph_drawing/till_october_2018/transfert_few_data/vgg_svhn_from_cifar100_deepstrom_few_data.py b/main/experiments/graph_drawing/till_october_2018/transfert_few_data/vgg_svhn_from_cifar100_deepstrom_few_data.py
new file mode 100644
index 0000000000000000000000000000000000000000..c40386f2524bc613efd1c0b96129685ba4fb7af7
--- /dev/null
+++ b/main/experiments/graph_drawing/till_october_2018/transfert_few_data/vgg_svhn_from_cifar100_deepstrom_few_data.py
@@ -0,0 +1,271 @@
+import os
+
+import matplotlib
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+import pathlib
+from skluc.main.utils import logger
+
+matplotlib.rcParams.update({'font.size': 14})
+
+pd.set_option('display.width', 1000)
+
+DAT = ["SVHN"]
+DIR = ["/home/luc/Resultats/Deepstrom/SVHN/june_2018/svhn_few_data_debug_plus_dropout"]
+
+
+for h in range(len(DIR)):
+ DATANAME = DAT[h]
+ DIRNAME = DIR[h]
+
+ FILENAME = "gathered_results.csv"
+
+ min_acc = 0.00
+ max_acc = 1.05
+ # max_acc = 1.0
+ linewidth = 0.9
+ output_conv_dim = 512
+ nb_classes = 10
+
+ real_nys_marker = "s"
+
+ learned_nys_marker = "x"
+
+ linearity_color = "g"
+
+ dense_marker = "v"
+ dense_color = "r"
+
+ deepfried_marker = "8"
+ deepfried_color = "b"
+
+ d_translate_kernel = {
+ "linear": "Linear",
+ "chi2_cpd": "Chi2",
+ "rbf": "Gaussian"
+ }
+
+ if __name__ == '__main__':
+ filepath = os.path.join(DIRNAME, FILENAME)
+ field_names = ["method_name",
+ "accuracy_val",
+ "accuracy_test",
+ "runtime",
+ "number_epoch",
+ "batch_size",
+ "repr_dim",
+ "two_layers_dense",
+ "kernel_deepstrom",
+ "gamma_kernel",
+ "constante_sigmoid",
+ "nb_layer_deepfried",
+ "subsample_size",
+ "validation_size",
+ "seed",
+ "non_linearity",
+ "real_nystrom",
+ "repr_quality",
+ "train_size",
+ "dropout"
+ ]
+
+ df = pd.read_csv(filepath, names=field_names)
+ # df = df[df["accuracy_val"] != 'None']
+ df = df.apply(pd.to_numeric, errors="ignore")
+ method_names = set(df["method_name"].values)
+ kernel_names = set(df["kernel_deepstrom"].values)
+ kernel_names.remove("None")
+ # kernel_names.remove("laplacian")
+ repr_dim = set(df["repr_dim"].values)
+ repr_dim.remove("None") # dtype: str
+ # repr_dim.remove("16")
+ nys_size = set(df["subsample_size"].values)
+ nys_size.remove("None")
+ nb_layers_deepfried = set(df["nb_layer_deepfried"].values)
+ nb_layers_deepfried.remove("None")
+ seed_values = set(df["seed"].values)
+ batch_size = 128
+ train_sizes = set(df["train_size"])
+ dropout_values = set(df["dropout"].values)
+ sigma_values = set(df["gamma_kernel"].values)
+ sigma_values.remove("None")
+ sigma_values = list(sigma_values)
+ logger.debug("Nystrom possible sizes are: {}".format(nys_size))
+ logger.debug("Kernel functions are: {}".format(kernel_names))
+ logger.debug("Compared network types are: {}".format(method_names))
+ logger.debug("Tested representation dimension are: {}".format(repr_dim))
+
+ means_deepstrom = {}
+
+ for t_size in sorted(list(train_sizes)):
+ df_tsize = df[df["train_size"] == t_size]
+ for drop_val in dropout_values:
+ if int(drop_val) != 1:
+ continue
+ df_drop = df_tsize[df_tsize["dropout"] == drop_val]
+
+ # plot deepstrom
+ # ==============
+ df_deepstrom = df_drop[df_drop["method_name"] == "deepstrom"]
+ df_deepstrom["subsample_size"] = df_deepstrom["subsample_size"].astype(np.int)
+ df_deepstrom_sort = df_deepstrom.sort_values(by=["subsample_size"])
+ for i, k_name in enumerate(sorted(kernel_names)):
+ if k_name != "rbf":
+ df_deepstrom_kernels = [df_deepstrom_sort[df_deepstrom_sort["kernel_deepstrom"] == k_name]]
+ else:
+ df_deepstrom_kernels = []
+ df_deepstrom_kernel_tmp = df_deepstrom_sort[df_deepstrom_sort["kernel_deepstrom"] == k_name]
+ for sig_val in sigma_values:
+ if float(sig_val) != 0.1:
+ continue
+ df_deepstrom_kernels.append(
+ df_deepstrom_kernel_tmp[df_deepstrom_kernel_tmp["gamma_kernel"] == sig_val])
+
+ for j, df_deepstrom_kernel in enumerate(df_deepstrom_kernels):
+ f, ax = plt.subplots()
+ # non_lin_dfs = {
+ # "linear": df_deepstrom_kernel[df_deepstrom_kernel["non_linearity"] == "None"],
+ # }
+ # get the results of learned nystrom
+ df_deepstrom_kernel_w = df_deepstrom_kernel[df_deepstrom_kernel["real_nystrom"] == False]
+ np_deepstrom_kernel_w_mean_accuracy_test = np.mean(np.array([
+ list(df_deepstrom_kernel_w[df_deepstrom_kernel_w["seed"] == seed_v]["accuracy_test"]) for
+ seed_v in seed_values
+ ]), axis=0)
+ np_deepstrom_kernel_w_std_accuracy_test = np.std(np.array(
+ [list(df_deepstrom_kernel_w[df_deepstrom_kernel_w["seed"] == seed_v]["accuracy_test"]) for
+ seed_v in
+ seed_values]), axis=0)
+ np_param_nbr_deepstrom_kernel_w = (
+ np.square(np.array(sorted(set(df_deepstrom_kernel_w["subsample_size"])))) + # m x m
+ np.array(
+ sorted(set(df_deepstrom_kernel_w["subsample_size"]))) * output_conv_dim + # m x d
+ np.array(
+ sorted(list(set(df_deepstrom_kernel_w["subsample_size"])))) * nb_classes) # m x c
+
+ ax.errorbar(np_param_nbr_deepstrom_kernel_w,
+ np_deepstrom_kernel_w_mean_accuracy_test,
+ np_deepstrom_kernel_w_std_accuracy_test,
+ marker=learned_nys_marker, color=linearity_color,
+ label="Adaptative Deepström",
+ capsize=3)
+
+ # get the results of vanilla nystrom
+ df_deepstrom_kernel_k = df_deepstrom_kernel[df_deepstrom_kernel["real_nystrom"]]
+ if len(df_deepstrom_kernel_k):
+ np_deepstrom_kernel_k_mean_accuracy_test = np.mean(
+ np.array([list(
+ df_deepstrom_kernel_k[df_deepstrom_kernel_k["seed"] == seed_v]["accuracy_test"]) for
+ seed_v in
+ seed_values]), axis=0)
+ np_deepstrom_kernel_k_std_accuracy_test = np.std(
+ np.array([list(
+ df_deepstrom_kernel_k[df_deepstrom_kernel_k["seed"] == seed_v]["accuracy_test"]) for
+ seed_v in
+ seed_values]), axis=0)
+
+ np_param_nbr_deepstrom_kernel_k = (
+ np.square(np.array(sorted(set(df_deepstrom_kernel_k["subsample_size"])))) + # m x m
+ np.array(sorted(
+ set(df_deepstrom_kernel_k["subsample_size"]))) * output_conv_dim + # m x d
+ np.array(sorted(
+ list(set(df_deepstrom_kernel_k["subsample_size"])))) * nb_classes) # m x c
+
+ ax.errorbar(np_param_nbr_deepstrom_kernel_k,
+ np_deepstrom_kernel_k_mean_accuracy_test,
+ np_deepstrom_kernel_k_std_accuracy_test,
+ marker=real_nys_marker, color=linearity_color,
+ label="Deepström",
+ capsize=3)
+
+ # plot dense
+ # ==========
+ df_dense = df_drop[df_drop["method_name"] == "dense"]
+ df_dense = df_dense[df_dense["train_size"] == t_size]
+ df_dense["repr_dim"] = df_dense["repr_dim"].astype(np.int)
+ df_dense = df_dense.sort_values(by=["repr_dim"])
+ np_dense_mean_accuracy_test = np.mean(
+ np.array([list(df_dense[df_dense["seed"] == seed_v]["accuracy_test"]) for seed_v in
+ seed_values]), axis=0)
+ np_dense_std_accuracy_test = np.std(
+ np.array([list(df_dense[df_dense["seed"] == seed_v]["accuracy_test"]) for seed_v in
+ seed_values]), axis=0)
+ ax.errorbar(
+ np.array(sorted([int(n) for n in np.unique(df_dense["repr_dim"])])) * output_conv_dim +
+ np.array(sorted([int(n) for n in np.unique(df_dense["repr_dim"])])) * nb_classes,
+ np_dense_mean_accuracy_test,
+ np_dense_std_accuracy_test,
+ color=dense_color,
+ marker=dense_marker,
+ label="Fully Connected", capsize=3)
+
+ # plot deepfried
+ # ==============
+ df_deepfried = df_drop[df_drop["method_name"] == "deepfriedconvnet"]
+ np_deepfried_mean_accuracy_test = []
+ np_deepfried_std_accuracy_test = []
+ for l_nb in sorted(nb_layers_deepfried):
+ df_deepfried_stack = df_deepfried[df_deepfried["nb_layer_deepfried"] == l_nb]
+ np_deepfried_mean_accuracy_test.append(np.mean(df_deepfried_stack["accuracy_test"]))
+ np_deepfried_std_accuracy_test.append(np.std(df_deepfried_stack["accuracy_test"]))
+
+ ax.errorbar([(output_conv_dim * 3 + output_conv_dim * nb_classes) * i for i in [1]],
+ np_deepfried_mean_accuracy_test,
+ np_deepfried_std_accuracy_test,
+ color=deepfried_color,
+ marker=deepfried_marker,
+
+ label="Adaptative DeepFriedConvnet", capsize=3)
+ ax.set_ylim(min_acc, max_acc)
+ ax.set_ylabel("Accuracy")
+ ax.set_xticks([1e4, 1e5, 1e6])
+ # if i == 2:
+ # ax.set_xlabel("# Parameters")
+ ax.set_xlabel("# Parameters")
+ ax.legend(bbox_to_anchor=(0.5, -0.20), loc="upper center", ncol=2)
+ ax.set_xticklabels([1e4, 1e5, 1e6])
+ # else:
+ # ax.set_xticklabels([])
+ ax.set_xscale("symlog")
+
+ ax_twin = ax.twiny()
+ ax_twin.set_xscale("symlog")
+ ax_twin.set_xlim(ax.get_xlim())
+ ax_twin.set_xticks(np_param_nbr_deepstrom_kernel_w)
+
+ # if i == 0:
+ ax_twin.set_xlabel("Subsample Size")
+ ax.set_title(
+ "{} Kernel - {} - Train size: {}".format(d_translate_kernel[k_name], DATANAME, t_size),
+ y=1.2)
+ ax_twin.set_xticklabels(sorted(set(df_deepstrom_kernel_w["subsample_size"])))
+ # else:
+ # ax.set_title("Noyau {} - {} - Train size: {}".format(d_translate_kernel[k_name], DATANAME, t_size))
+ # ax_twin.set_xticklabels([])
+
+ f.set_size_inches(8, 6)
+ f.tight_layout()
+ f.subplots_adjust(bottom=0.3)
+ # f.show()
+ # exit()
+ # learnable: change legend
+ # ODIR = [
+ # "/home/luc/PycharmProjects/deepFriedConvnets/main/experiments/graph_drawing/paper/svhn/few_data/parameters/dropout_{}".format(
+ # str(drop_val).replace(".", "-"))]
+ # out_dir_path = ODIR[h]
+
+
+ if k_name != "rbf":
+ out_name = "acc_param_tsize_{}_{}_{}".format(t_size, str(drop_val).replace(".", "-"),
+ k_name)
+ else:
+ out_name = "acc_param_tsize_{}_{}_{}_{}".format(t_size, str(drop_val).replace(".", "-"),
+ k_name,
+ str(sigma_values[j]).replace(".", "-"))
+
+ base_out_dir = os.path.join(os.path.abspath(__file__.split(".")[0]), "images")
+ pathlib.Path(base_out_dir).mkdir(parents=True, exist_ok=True)
+ out_path = os.path.join(base_out_dir, out_name)
+ logger.debug(out_path)
+ f.savefig(out_path)
diff --git a/main/experiments/graph_drawing/till_october_2018/transfert_few_data_cifar100_from_cifar10/vgg_deepstrom_few_data_cifar100_from_cifar10.py b/main/experiments/graph_drawing/till_october_2018/transfert_few_data_cifar100_from_cifar10/vgg_deepstrom_few_data_cifar100_from_cifar10.py
new file mode 100644
index 0000000000000000000000000000000000000000..978978b94b39a2c7bdeb2509350692fed240b60e
--- /dev/null
+++ b/main/experiments/graph_drawing/till_october_2018/transfert_few_data_cifar100_from_cifar10/vgg_deepstrom_few_data_cifar100_from_cifar10.py
@@ -0,0 +1,254 @@
+import os
+
+import matplotlib
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+import pathlib
+from skluc.main.utils import logger
+
+matplotlib.rcParams.update({'font.size': 14})
+
+# pd.set_option('display.width', 1000)
+pd.set_option('display.expand_frame_repr', False)
+
+# DAT = ["SVHN"]
+# DIR = ["/home/luc/Resultats/Deepstrom/october_2018/transfert_few_data"]
+
+
+DATANAME = "SVHN"
+DIRNAME = "/home/luc/Resultats/Deepstrom/october_2018/transfert_few_data"
+
+FILENAME = "gathered_results_all.csv"
+
+min_acc = 0.00
+max_acc = 1.05
+# max_acc = 1.0
+linewidth = 0.9
+output_conv_dim = 512
+nb_classes = 10
+
+real_nys_marker = "s"
+
+learned_nys_marker = "x"
+
+linearity_color = "g"
+
+dense_marker = "v"
+dense_color = "r"
+
+deepfried_marker = "8"
+deepfried_color = "b"
+
+d_translate_kernel = {
+ "linear": "Linear",
+ "chi2_cpd": "Chi2",
+ "rbf": "Gaussian"
+}
+
+if __name__ == '__main__':
+ filepath = os.path.join(DIRNAME, FILENAME)
+ field_names = ["method_name",
+ "accuracy_val",
+ "accuracy_test",
+ "runtime_train",
+ "runtime_val",
+ "runtime_test",
+ "number_epoch",
+ "batch_size",
+ "repr_dim",
+ "second_layer_size",
+ "kernel_deepstrom",
+ "gamma_kernel",
+ "constante_sigmoid",
+ "nb_layer_deepfried",
+ "subsample_size",
+ "validation_size",
+ "seed",
+ "act",
+ "non_linearity",
+ "real_nystrom",
+ "repr_quality",
+ "train_size",
+ "dropout",
+ "dataset",
+ "real_deepfried"
+ ]
+
+ df = pd.read_csv(filepath, names=field_names)
+ df = df[df["accuracy_val"] != 'None']
+ df = df.apply(pd.to_numeric, errors="ignore")
+ df = df.drop_duplicates()
+ method_names = set(df["method_name"].values)
+ kernel_names = set(df["kernel_deepstrom"].values)
+ kernel_names.remove("None")
+ # kernel_names.remove("laplacian")
+ repr_dim = set(df["repr_dim"].values)
+ repr_dim.remove("None") # dtype: str
+ # repr_dim.remove("16")
+ nys_size = set(df["subsample_size"].values)
+ nys_size.remove("None")
+ nb_layers_deepfried = set(df["nb_layer_deepfried"].values)
+ nb_layers_deepfried.remove("None")
+ seed_values = set(df["seed"].values)
+ batch_size = 128
+ train_sizes = set(df["train_size"])
+
+ cut_layers = set(df["repr_quality"].values)
+
+ logger.debug("Nystrom possible sizes are: {}".format(nys_size))
+ logger.debug("Kernel functions are: {}".format(kernel_names))
+ logger.debug("Compared network types are: {}".format(method_names))
+ logger.debug("Tested representation dimension are: {}".format(repr_dim))
+
+ means_deepstrom = {}
+
+ for t_size in sorted(list(train_sizes)):
+ df_tsize = df[df["train_size"] == t_size]
+
+ for cut_layer in cut_layers:
+ df_cut_layer = df_tsize[df_tsize["repr_quality"] == cut_layer]
+
+ # plot deepstrom
+ # ==============
+ df_deepstrom = df_cut_layer[df_cut_layer["method_name"] == "deepstrom"]
+ df_deepstrom["subsample_size"] = df_deepstrom["subsample_size"].astype(np.int)
+ df_deepstrom_sort = df_deepstrom.sort_values(by=["subsample_size"])
+ for k_name in sorted(kernel_names):
+ df_deepstrom_kernel = df_deepstrom_sort[df_deepstrom_sort["kernel_deepstrom"] == k_name]
+
+ f, ax = plt.subplots()
+
+ # get the results of learned nystrom
+ df_deepstrom_kernel_w = df_deepstrom_kernel[df_deepstrom_kernel["real_nystrom"] == False]
+ all_accs_w = np.array([
+ list(df_deepstrom_kernel_w[df_deepstrom_kernel_w["seed"] == seed_v]["accuracy_test"]) for
+ seed_v in seed_values
+ ])
+ np_deepstrom_kernel_w_mean_accuracy_test = np.mean(all_accs_w, axis=0)
+ np_deepstrom_kernel_w_std_accuracy_test = np.std(all_accs_w, axis=0)
+ np_param_nbr_deepstrom_kernel_w = (
+ np.square(np.array(sorted(set(df_deepstrom_kernel_w["subsample_size"])))) + # m x m
+ np.array(
+ sorted(set(df_deepstrom_kernel_w["subsample_size"]))) * output_conv_dim + # m x d
+ np.array(
+ sorted(list(set(df_deepstrom_kernel_w["subsample_size"])))) * nb_classes) # m x c
+
+ ax.errorbar(np_param_nbr_deepstrom_kernel_w,
+ np_deepstrom_kernel_w_mean_accuracy_test,
+ np_deepstrom_kernel_w_std_accuracy_test,
+ marker=learned_nys_marker, color=linearity_color,
+ label="Adaptative Deepström",
+ capsize=3)
+
+ # get the results of vanilla nystrom
+ df_deepstrom_kernel_k = df_deepstrom_kernel[df_deepstrom_kernel["real_nystrom"]]
+ if len(df_deepstrom_kernel_k):
+ all_accs_k = np.array([
+ list(df_deepstrom_kernel_k[df_deepstrom_kernel_k["seed"] == seed_v]["accuracy_test"]) for
+ seed_v in seed_values
+ ])
+ np_deepstrom_kernel_k_mean_accuracy_test = np.mean(all_accs_k, axis=0)
+ np_deepstrom_kernel_k_std_accuracy_test = np.std(all_accs_k, axis=0)
+
+ np_param_nbr_deepstrom_kernel_k = (
+ np.square(np.array(sorted(set(df_deepstrom_kernel_k["subsample_size"])))) + # m x m
+ np.array(sorted(
+ set(df_deepstrom_kernel_k["subsample_size"]))) * output_conv_dim + # m x d
+ np.array(sorted(
+ list(set(df_deepstrom_kernel_k["subsample_size"])))) * nb_classes) # m x c
+
+ ax.errorbar(np_param_nbr_deepstrom_kernel_k,
+ np_deepstrom_kernel_k_mean_accuracy_test,
+ np_deepstrom_kernel_k_std_accuracy_test,
+ marker=real_nys_marker, color=linearity_color,
+ label="Deepström",
+ capsize=3)
+
+ # plot dense
+ # ==========
+ df_dense = df_cut_layer[df_cut_layer["method_name"] == "dense"]
+ df_dense = df_dense[df_dense["train_size"] == t_size]
+ df_dense["repr_dim"] = df_dense["repr_dim"].astype(np.int)
+ df_dense = df_dense.sort_values(by=["repr_dim"])
+ np_dense_mean_accuracy_test = np.mean(
+ np.array([list(df_dense[df_dense["seed"] == seed_v]["accuracy_test"]) for seed_v in
+ seed_values]), axis=0)
+ np_dense_std_accuracy_test = np.std(
+ np.array([list(df_dense[df_dense["seed"] == seed_v]["accuracy_test"]) for seed_v in
+ seed_values]), axis=0)
+ ax.errorbar(
+ np.array(sorted([int(n) for n in np.unique(df_dense["repr_dim"])])) * output_conv_dim +
+ np.array(sorted([int(n) for n in np.unique(df_dense["repr_dim"])])) * nb_classes,
+ np_dense_mean_accuracy_test,
+ np_dense_std_accuracy_test,
+ color=dense_color,
+ marker=dense_marker,
+ label="Fully Connected", capsize=3)
+
+ # # plot deepfried
+ # # ==============
+ df_deepfried = df_cut_layer[df_cut_layer["method_name"] == "deepfriedconvnet"]
+ np_deepfried_mean_accuracy_test = []
+ np_deepfried_std_accuracy_test = []
+ for l_nb in sorted(nb_layers_deepfried):
+ df_deepfried_stack = df_deepfried[df_deepfried["nb_layer_deepfried"] == l_nb]
+ if len(df_deepfried_stack):
+ np_deepfried_mean_accuracy_test.append(np.mean(df_deepfried_stack["accuracy_test"]))
+ np_deepfried_std_accuracy_test.append(np.std(df_deepfried_stack["accuracy_test"]))
+
+ nb_param_vals = [(output_conv_dim * 3 + output_conv_dim * nb_classes) * int(i) for i in sorted(set(df_deepfried["nb_layer_deepfried"].values))]
+ ax.errorbar(nb_param_vals,
+ np_deepfried_mean_accuracy_test,
+ np_deepfried_std_accuracy_test,
+ color=deepfried_color,
+ marker=deepfried_marker,
+ label="Adaptative DeepFriedConvnet", capsize=3)
+
+
+ ax.set_ylim(min_acc, max_acc)
+ ax.set_ylabel("Accuracy")
+ ax.set_xticks([1e4, 1e5, 1e6])
+ # if i == 2:
+ # ax.set_xlabel("# Parameters")
+ ax.set_xlabel("# Parameters")
+ ax.legend(bbox_to_anchor=(0.5, -0.20), loc="upper center", ncol=2)
+ ax.set_xticklabels([1e4, 1e5, 1e6])
+ # else:
+ # ax.set_xticklabels([])
+ ax.set_xscale("symlog")
+
+ ax_twin = ax.twiny()
+ ax_twin.set_xscale("symlog")
+ ax_twin.set_xlim(ax.get_xlim())
+ ax_twin.set_xticks(np_param_nbr_deepstrom_kernel_w)
+
+ # if i == 0:
+ ax_twin.set_xlabel("Subsample Size")
+ ax.set_title(
+ "{} Kernel - {} - Train size: {}".format(d_translate_kernel[k_name], DATANAME, t_size),
+ y=1.2)
+ ax_twin.set_xticklabels(sorted(set(df_deepstrom_kernel_w["subsample_size"])))
+ # else:
+ # ax.set_title("Noyau {} - {} - Train size: {}".format(d_translate_kernel[k_name], DATANAME, t_size))
+ # ax_twin.set_xticklabels([])
+
+ f.set_size_inches(8, 6)
+ f.tight_layout()
+ f.subplots_adjust(bottom=0.3)
+ # f.show()
+ # exit()
+ # learnable: change legend
+ # ODIR = [
+ # "/home/luc/PycharmProjects/deepFriedConvnets/main/experiments/graph_drawing/paper/svhn/few_data/parameters/dropout_{}".format(
+ # str(drop_val).replace(".", "-"))]
+ # out_dir_path = ODIR[h]
+
+
+ out_name = "acc_param_tsize_{}_{}_{}".format(t_size, cut_layer, k_name)
+
+ base_out_dir = os.path.join(os.path.abspath(__file__.split(".")[0]), "images")
+ pathlib.Path(base_out_dir).mkdir(parents=True, exist_ok=True)
+ out_path = os.path.join(base_out_dir, out_name)
+ logger.debug(out_path)
+ f.savefig(out_path)
diff --git a/main/experiments/graph_drawing/vgg_cifar_deepstrom_big_subsample.py b/main/experiments/graph_drawing/till_october_2018/vgg_cifar_deepstrom_big_subsample.py
similarity index 100%
rename from main/experiments/graph_drawing/vgg_cifar_deepstrom_big_subsample.py
rename to main/experiments/graph_drawing/till_october_2018/vgg_cifar_deepstrom_big_subsample.py
diff --git a/main/experiments/graph_drawing/vgg_cifar_deepstrom_little_subsample.py b/main/experiments/graph_drawing/till_october_2018/vgg_cifar_deepstrom_little_subsample.py
similarity index 100%
rename from main/experiments/graph_drawing/vgg_cifar_deepstrom_little_subsample.py
rename to main/experiments/graph_drawing/till_october_2018/vgg_cifar_deepstrom_little_subsample.py
diff --git a/main/experiments/graph_drawing/vgg_cifar_deepstrom_relu.py b/main/experiments/graph_drawing/till_october_2018/vgg_cifar_deepstrom_relu.py
similarity index 100%
rename from main/experiments/graph_drawing/vgg_cifar_deepstrom_relu.py
rename to main/experiments/graph_drawing/till_october_2018/vgg_cifar_deepstrom_relu.py
diff --git a/main/experiments/graph_drawing/vgg_cifar_deepstrom_runtime.py b/main/experiments/graph_drawing/till_october_2018/vgg_cifar_deepstrom_runtime.py
similarity index 100%
rename from main/experiments/graph_drawing/vgg_cifar_deepstrom_runtime.py
rename to main/experiments/graph_drawing/till_october_2018/vgg_cifar_deepstrom_runtime.py
diff --git a/main/experiments/parameter_files/october_2018/lazyfile_transfert_few_data.yml b/main/experiments/parameter_files/october_2018/lazyfile_transfert_few_data.yml
new file mode 100644
index 0000000000000000000000000000000000000000..2224de04489bc0d546da6f22f2de037266eced93
--- /dev/null
+++ b/main/experiments/parameter_files/october_2018/lazyfile_transfert_few_data.yml
@@ -0,0 +1,61 @@
+all:
+ dense:
+ deepfried:
+ deepstrom_real_gamma:
+ deepstrom_real_no_kernel_param:
+ deepstrom_learned_gamma:
+ deepstrom_learned_no_kernel_param:
+
+base:
+ epoch_numbers: {"-e": [50]}
+ batch_sizes: {"-s": [64]}
+ val_size: {"-v": [10000]}
+ seed: {"-a": "range(5)"}
+ dropout: {"-d": [0.5, 0.7, 1.0]}
+ data_size: {"-t":[20, 50, 100, 200, 500, 1000, 2000]}
+ dataset: ["--svhn"]
+
+gammavalue:
+ gamma: {"-g": [0.0001, 0.001, 0.0025145440260884045, 0.01, 0.1]}
+
+dense:
+ network: ["dense"]
+ base:
+ repr_dim: {"-D": [16, 64, 128, 1024]}
+
+deepfried:
+ network: ["deepfriedconvnet"]
+ base:
+ gammavalue:
+ nbstacks: {"-N": [1]}
+
+deepstrom_base:
+ network: ["deepstrom"]
+ base:
+ nys_size: {"-m": [16, 64, 128, 256, 512, 1024]}
+
+deepstrom_real:
+ deepstrom_base:
+ real_nys: ["-r"]
+
+deepstrom_real_gamma:
+ deepstrom_real:
+ gammavalue:
+ kernel: ["-R"]
+
+deepstrom_real_no_kernel_param:
+ deepstrom_real:
+ kernel: ["-C", "-L"]
+
+deepstrom_learned:
+ deepstrom_base:
+
+deepstrom_learned_gamma:
+ deepstrom_learned:
+ gammavalue:
+ kernel: ["-R"]
+
+deepstrom_learned_no_kernel_param:
+ deepstrom_learned:
+ kernel: ["-C", "-L"]
+
diff --git a/main/experiments/parameter_files/october_2018/lazyfile_transfert_few_data_cifar100_from_cifar10.yml b/main/experiments/parameter_files/october_2018/lazyfile_transfert_few_data_cifar100_from_cifar10.yml
new file mode 100644
index 0000000000000000000000000000000000000000..4dc38842bd4650815b3e1df3502179acc7ec1b93
--- /dev/null
+++ b/main/experiments/parameter_files/october_2018/lazyfile_transfert_few_data_cifar100_from_cifar10.yml
@@ -0,0 +1,33 @@
+all:
+ dense:
+ deepfried:
+ deepstrom:
+
+base:
+ epoch_numbers: {"-e": [100]}
+ batch_sizes: {"-s": [64]}
+ val_size: {"-v": [10000]}
+ seed: {"-a": "range(10)"}
+ quiet: ["-q"]
+ data_size: {"-t":[20, 50, 100, 200, 500, 1000, 2000]}
+ dataset: ["--svhn"]
+ weights: {"-W": ["cifar100"]}
+ cut_layer: {"-B": ["block3_pool", "block5_conv4", "block5_pool"]}
+
+dense:
+ network: ["dense"]
+ base:
+ repr_dim: {"-D": [16, 64, 128, 1024]}
+
+deepfried:
+ network: ["deepfriedconvnet"]
+ base:
+ nbstacks: {"-N": [1, 3, 5, 7]}
+
+deepstrom:
+ network: ["deepstrom"]
+ base:
+ real_nys: ["-r", ""]
+ nys_size: {"-m": [16, 64, 128, 256, 512, 1024]}
+ kernel: ["-C", "-L", "-R"]
+