Aliases:
tf.nn.local_response_normalization
tf.nn.lrn
tf.nn.local_response_normalization(
input,
depth_radius=5,
bias=1,
alpha=1,
beta=0.5,
name=None
)
Defined in generated file: tensorflow/python/ops/gen_nn_ops.py
.
Local Response Normalization.
The 4-D input
tensor is treated as a 3-D array of 1-D vectors (along the last
dimension), and each vector is normalized independently. Within a given vector,
each component is divided by the weighted, squared sum of inputs within
depth_radius
. In detail,
sqr_sum[a, b, c, d] =
sum(input[a, b, c, d - depth_radius : d + depth_radius + 1] ** 2)
output = input / (bias + alpha * sqr_sum) ** beta
For details, see Krizhevsky et al., ImageNet classification with deep convolutional neural networks (NIPS 2012).
Args:
input
: ATensor
. Must be one of the following types:half
,bfloat16
,float32
. 4-D.depth_radius
: An optionalint
. Defaults to5
. 0-D. Half-width of the 1-D normalization window.bias
: An optionalfloat
. Defaults to1
. An offset (usually positive to avoid dividing by 0).alpha
: An optionalfloat
. Defaults to1
. A scale factor, usually positive.beta
: An optionalfloat
. Defaults to0.5
. An exponent.name
: A name for the operation (optional).
Returns:
A Tensor
. Has the same type as input
.