Reduce a size of neural networks by at least removing 406 one or more neurons of the neural network and adjusting one or more layers of the neural network to compensate for the removed one or more neurons. The adjustment may include propagating one or more bias terms of the removed one or more neurons to a subsequent layer, this propagating may include calculating one or more activation function results of one or more bias terms. The removing may include identifying zero-weighted neurons in a representation of the neural network and removing them. The adjustment may include determining that removing one or more neurons would result in element-wise operation performed between layers having operands of different length, and inserting one or more scattered operations to generate operands of equal length for the element-wise operation.
Techniques for pruning neural networks
2022-08-03
Patent
Electronic Resource
English
IPC: | G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / B60W CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION , Gemeinsame Steuerung oder Regelung von Fahrzeug-Unteraggregaten verschiedenen Typs oder verschiedener Funktion |
ACTIVATION ZERO-BYPASS AND WEIGHT PRUNING IN NEURAL NETWORKS FOR VEHICLE PERCEPTION SYSTEMS
European Patent Office | 2020
|Activation zero-bypass and weight pruning in neural networks for vehicle perception systems
European Patent Office | 2022
|Catalogue agriculture | 2008
|