INFORMATION PROCESSING SYSTEM AND INFORMATION PROCESSING METHOD

Title:

INFORMATION PROCESSING SYSTEM AND INFORMATION PROCESSING METHOD

Document Type and Number:

WIPO Patent Application WO/2020/255634

Kind Code:

A1

Abstract:

This information processing system is provided with a learning unit (152) that carries out learning using a neural network, on the basis of data for learning. The neural network includes: a parameter-calculating network that calculates a predetermined parameter; and an optimizing layer that executes a predetermined optimizing calculation on the basis of the parameter calculated by the parameter-calculating network. The learning unit (152) updates the parameter of the neural network by backpropagating an error based on a result of the optimizing calculation executed by the optimizing layer.

More Like This:

WO/2023/245194	PARTITIONING INFORMATION IN NEURAL NETWORK-BASED VIDEO CODING
JP5207870	Dimension reduction method, pattern recognition dictionary generator, and pattern recognition device
JP2022113117	NEURAL NETWORK FOR DOMAIN ADAPTATION AND TRAINING METHOD THEREOF

Inventors:

IDE NAOKI (JP)

Application Number:

PCT/JP2020/020427

Publication Date:

December 24, 2020

Filing Date:

May 22, 2020

Export Citation:

Click for automatic bibliography generation Help

Assignee:

SONY CORP (JP)

International Classes:

G06N3/08; G06N10/00; G06N99/00

Foreign References:

JP2018156473A

2018-10-04

Other References:

OHZEKI MASAYUKI, OKADA SHUNTARO, TERABE MASAYOSHI, TAGUCHI SHINICHIRO: "Optimization of neural networks via finite- value quantum fluctuations", SCIENTIFIC REPORTS, vol. 8, no. 9950, 2 July 2018 (2018-07-02), pages 1 - 10, XP055773217, Retrieved from the Internet DOI: 10.1038/ s 41598-018 -28212-4

Attorney, Agent or Firm:

SAKAI INTERNATIONAL PATENT OFFICE (JP)

Download PDF:

View/Download PDF PDF Help

Previous Patent: GLASS TRANSFER DEVICE

Next Patent: WORK MACHINE, SYSTEM, AND METHOD FOR CONTROLLING WORK MACHINE