TensorSpace.js
Getting Start
Basic Concepts
TensorSpace Converter
Model Preprocessing
Models
Layers
Merge Function
Merge Function Introduction
"Merge" functions are used to provide a way to present different "layer merging" operations among different layers. There are some key points we want to highlight here for the "Merge" functions.
Factory function
We designed all merge function as a "Factory" for layer merging operation. The factory returns a merged layer object with providing some proper parameters: a list of layer objects need to be merged and some configurations of the returned merged object.
List of input layers
The list should contain layers need to be merged. The provided layers need to match the shape requirements based on the corresponding merge function. For example, all layers for "Add" function should have the same shape.
Output configurations
The configurations is used to describe the generated merged layer object. The configurations include the basic settings and status about the merged layer object.
Supported Merge Functions
filter_center_focus Add
filter_center_focus Average
filter_center_focus Maximum
filter_center_focus Multiply
filter_center_focus Subtract
filter_center_focus Concatenate