iframe-proxy | Sunbelt Computer Software

Repository navigation

import math

import numpy as np

import tensorflow as tf

def lrelu(x, leak=0.2, name="activation"):

with tf.variable_scope(name):

f = tf.maximum(x, leak*x, name='f')

df = tf.truediv(f, x, name='df')

return f

def fc(input_, output_size, fn=None, stddev=0.02, bias_start=0.0, name=None):

"""Fully-Connected Layer

"""

shape = input_.get_shape().as_list()

with tf.variable_scope(name):

# Define Weights and Biases in layer scope

weight = tf.get_variable("weight", [shape[1], output_size], tf.float32,

tf.random_normal_initializer(stddev=stddev))

bias = tf.get_variable("bias", [output_size],

initializer=tf.constant_initializer(bias_start))

act = tf.matmul(input_, weight) + bias

# Apply activation function fn

if fn is not None:

out = fn(act)

else:

out = act

return out