Developer Guide for Intel® Data Analytics Acceleration Library 2019 Update 1
For a two-dimensional input tensor X ∈ R n 1 x 1 with batch dimension of size n 1 and two-dimensional T ∈ R n 1 x 1, the layer computes a one-dimensional tensor with the logistic cross-entropy value. For more details, see Forward Loss Logistic Cross-entropy Layer.
The backward loss logistic cross-entropy layer for a given two-dimensional tensor X ∈ R n 1 x 1 and two-dimensional T ∈ R n 1 x 1 computes the two-dimensional tensor Z ∈ R n 1 x 1 such that
where
pr i ∈ [0,1] is the probability that a sample belongs to the first of two classes