base_ops.py

import keras.backend as K

def switch(condition, t, e):
    if K.backend() == 'tensorflow':
        import tensorflow as tf
        return tf.where(condition, t, e)
    elif K.backend() == 'theano':
        import theano.tensor as tt
        return tt.switch(condition, t, e)


def round_through(x):
    '''Element-wise rounding to the closest integer with full gradient propagation.
    A trick from [Sergey Ioffe](http://stackoverflow.com/a/36480182)
    '''
    rounded = K.round(x)
    return x + K.stop_gradient(rounded - x)


def clip_through(x, min, max):
    '''Element-wise rounding to the closest integer with full gradient propagation.
    A trick from [Sergey Ioffe](http://stackoverflow.com/a/36480182)
    '''
    clipped = K.clip(x,min,max)
    return x + K.stop_gradient(clipped - x)

def _hard_sigmoid(x):
    '''Hard sigmoid different from the more conventional form (see definition of K.hard_sigmoid).

    # Reference:
    - [BinaryNet: Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1, Courbariaux et al. 2016](http://arxiv.org/abs/1602.02830}

    '''
    x = (0.5 * x) + 0.5
    return K.clip(x, 0, 1)

def log2_through(W):
    constant = np.log(2)
    logged = K.log(K.abs(W))/constant
    return W + K.stop_gradient(logged-W)


def pow_through(W, base):
    ones = K.ones_like(W)
    base = ones + base - 1
    powwed = K.pow(base, W)
    return W + K.stop_gradient(powwed - W)