dvgodoy

import numpy as np
from keras.models import Sequential
from keras.layers import Dense
from keras.optimizers import SGD
from keras.initializers import glorot_normal, normal

# ======================= #
# Data generation process #
# ======================= #

dvgodoy

from deepreplay.callbacks import ReplayData
from deepreplay.datasets.parabola import load_data

X, y = load_data()

replaydata = ReplayData(X, y, filename='hyperparms_in_action.h5', group_name='part1')

dvgodoy

from keras.models import Sequential
from keras.layers import Dense
from keras.optimizers import SGD
from keras.initializers import glorot_normal, normal

model = Sequential()
model.add(Dense(input_dim=2,
                units=2,
                activation='sigmoid',
                kernel_initializer=glorot_normal(seed=42),

dvgodoy

from deepreplay.replay import Replay

replay = Replay(replay_filename='hyperparms_in_action.h5', group_name='part1')

dvgodoy

import matplotlib.pyplot as plt

fig, ax = plt.subplots(1, 1, figsize=(5, 5))

fs = replay.build_feature_space(ax, layer_name='hidden')

dvgodoy

# Plot 60th epoch and save it as PNG
fs.plot(epoch=60).savefig('feature_space_epoch60.png', dpi=120)

# Animate and save it as MP4
fs.animate().save('feature_space_animation.mp4', dpi=120, fps=5)

dvgodoy

fig = plt.figure(figsize=(12, 6))
ax_fs = plt.subplot2grid((2, 4), (0, 0), colspan=2, rowspan=2)
ax_ph_neg = plt.subplot2grid((2, 4), (0, 2))
ax_ph_pos = plt.subplot2grid((2, 4), (1, 2))
ax_lm = plt.subplot2grid((2, 4), (0, 3))
ax_lh = plt.subplot2grid((2, 4), (1, 3))

fs = replay.build_feature_space(ax_fs, layer_name='hidden')
ph = replay.build_probability_histogram(ax_ph_neg, ax_ph_pos)
lh = replay.build_loss_histogram(ax_lh)

dvgodoy

from keras.models import Sequential
from keras.layers import Dense

def build_model(n_layers, input_dim, units, activation, initializer):
    if isinstance(units, list):
        assert len(units) == n_layers
    else:
        units = [units] * n_layers
        
    model = Sequential()

dvgodoy

from deepreplay.datasets.ball import load_data

X, y = load_data(n_dims=10)

dvgodoy

import numpy as np
from keras import backend as K
from keras.initializers import VarianceScaling

fan_in = fan_out = 100
stddev = np.sqrt(1. / fan_in)

normal_values = K.eval(K.random_normal(shape=(fan_in, fan_out), stddev=stddev)).ravel()
truncated_values = K.eval(K.truncated_normal(shape=(fan_in, fan_out), stddev=stddev)).ravel()
var_scaling_values = K.eval(VarianceScaling(mode='fan_in')(shape=(fan_in, fan_out))).ravel()