US Patent Application 18337462. Channel Gating For Conditional Computation simplified abstract
Contents
Channel Gating For Conditional Computation
Organization Name
Inventor(s)
Babak Ehteshami Bejnordi of Amsterdam (NL)
Tijmen Pieter Frederik Blankevoort of Amsterdam (NL)
Channel Gating For Conditional Computation - A simplified explanation of the abstract
- This abstract for appeared for US patent application number 18337462 Titled 'Channel Gating For Conditional Computation'
Simplified Explanation
This abstract describes a method for intelligently activating filters within a layer of a neural network. The method involves receiving input data specific to the layer, generating statistics based on this data, and assigning a relevance score to each filter in the layer. The relevance score indicates how relevant each filter is to the input data. Based on these relevance scores, the method determines the activation status of each filter and applies the input data to the activated filters to generate an output activation for the layer.
Original Abstract Submitted
A computing device may be configured to intelligently activate gating within a current layer of a neural network that includes two or more filters. The computing device may receive a layer-specific input data that is specific to the current layer of the neural network, generate statistics based on the received layer-specific input data; and use the generated statistics to assign a relevance score to each of the two or more filters. Each assigned relevance score may indicate the relevance of the corresponding filter to the received layer-specific input data. The computing device may determine an activation status of each of the two or more filters in the current layer based on the identified relevance and apply the received layer-specific input data to the activated filters in the two or more filters to generate an output activation for the current layer of the neural network.
