Skip to content

Latest commit

 

History

History
38 lines (23 loc) · 1.36 KB

MatrixMultiplicationLayer.md

File metadata and controls

38 lines (23 loc) · 1.36 KB

CMatrixMultiplicationLayer Class

This class implements a layer that performs matrix multiplication operation for its two inputs.

It treats the first three dimensions (BatchLength * BatchWidth * ListSize) as the number of matrices, the next three (Height * Width * Depth) as the height of each and the last one (Channels) as the width of each matrix. So this layer implements a bunch of independent matrix multiplications.

Settings

This layer has no settings.

Trainable parameters

This layer has no trainable parameters.

Inputs

This layer has 2 inputs. Each of them accepts the set of matrices:

  • BatchLength * BatchWidth * ListSize - the number of matrices in sets, must be equal between inputs
  • Height * Width * Depth - matrix height
  • Channels - matrix width; Channels of the first input must be equal to Height * Width * Depth of the second input

Outputs

Single output returns the blob with multiplication results, of the dimensions:

  • BatchLength, BatchWidth, ListSize, Height, Width, Depth are equal to these dimensions of the first input
  • Channels is equal to the Channels of the second input.