mxnet
|
support for unpool More...
Go to the source code of this file.
Classes | |
struct | mshadow::expr::UnPoolingExp< Reducer, SrcExp, DType, srcdim > |
unpooling expr reverse operation of pooling, used to pass gradient back More... | |
struct | mshadow::expr::Plan< UnPoolingExp< Reducer, SrcExp, DType, srcdim >, DType > |
Namespaces | |
mshadow | |
namespace for mshadow | |
mshadow::expr | |
namespace for abstract expressions and expressions template, have no dependency on tensor.h, These data structure takes no charge in computations, they are only used to define operations and represent expression in a symbolic way | |
Functions | |
template<typename Reducer , typename SrcExp , typename DType , int etype> | |
UnPoolingExp< Reducer, SrcExp, DType, ExpInfo< SrcExp >::kDim > | mshadow::expr::unpool (const Exp< SrcExp, DType, etype > &data_src, const Exp< SrcExp, DType, etype > &data_pooled, const Exp< SrcExp, DType, etype > &grad_pooled, index_t ksize_y, index_t ksize_x, index_t kstride_y, index_t kstride_x) |
unpooling gradient for 4D, backprop gradient value back, revserse operation of pooling, same as unpooling, but allows unequal size of kernel More... | |
support for unpool
Copyright (c) 2014 by Contributors