Search International and National Patent Collections

1. (WO2018220957) CONVOLUTIONAL NEURAL NETWORK

Pub. No.:    WO/2018/220957    International Application No.:    PCT/JP2018/011272
Publication Date: Fri Dec 07 00:59:59 CET 2018 International Filing Date: Fri Mar 23 00:59:59 CET 2018
IPC: G06N 3/063
G06G 7/60
Applicants: DENSO CORPORATION
株式会社デンソー
Inventors: KATAEVA Irina
カタエヴァ イリナ
Title: CONVOLUTIONAL NEURAL NETWORK
Abstract:
A neural network (20) is provided with a plurality of convolution layers (21-27) and a merging layer (63). One convolution layer has a crossbar circuit (44) that has a plurality of input bars (50), a plurality of output bars (51, 52), and a plurality of weight assignment elements (53) which assign weights. The crossbar circuit (44) carries out convolution calculation in an analog region by assigning weights to input signals and adding the input signals together on each of the output bars. Input data includes a plurality of feature maps. The crossbar circuit (44) has a first crossbar circuit (61) that carries out convolution calculation for a portion of the feature maps, and a second crossbar circuit (62) that carries out convolution calculation for another portion of the feature maps. The merging layer (63) merges the calculation results of the first and second crossbar circuits.