Title:
NEURAL NETWORK WEIGHT MATRIX ADJUSTMENT METHOD, WRITING CONTROL METHOD, AND RELATED DEVICE
Document Type and Number:
WIPO Patent Application WO/2021/163866
Kind Code:
A1
Abstract:
A neural network weight matrix adjustment method, a writing control method, and a related device, suitable for processing a neural network weight matrix before performing neural network operation by a computing-in-memory chip. The method comprises: determining whether weight distribution of a neural network weight matrix is lower than a first preset threshold; if yes, multiplying each weight value in the neural network weight matrix by a first constant; if not, determining whether the weight distribution of the neural network weight matrix is higher than a second preset threshold, wherein the second preset threshold is higher than the first preset threshold; and if the weight distribution of the neural network weight matrix is higher than the second preset threshold, dividing each weight value in the neural network weight matrix by a second constant, wherein the first constant and the second constant are both greater than 1. Thus, a signal obtained after computing-in-memory is performed in a memory cell array storing the processed weight matrix is within the effective range of an ADC. The computing accuracy is improved.
Inventors:
WANG SHAODI (CN)
Application Number:
PCT/CN2020/075648
Publication Date:
August 26, 2021
Filing Date:
February 18, 2020
Export Citation:
Assignee:
HANGZHOU ZHICUN INTELLIGENT TECH CO LTD (CN)
International Classes:
G06F9/30
Foreign References:
| CN105320495A | 2016-02-10 | |||
| CN109784485A | 2019-05-21 | |||
| CN107994570A | 2018-05-04 | |||
| CN109791626A | 2019-05-21 | |||
| CN110413255A | 2019-11-05 |
Attorney, Agent or Firm:
BEIJING SANYOU INTELLECTUAL PROPERTY AGENCY LTD. (CN)
Download PDF: