COMPRESSED DATA

To provide compression data which allows for highly compressing neural network parameter data.SOLUTION: Compressed data is generated by encoding quantization information defining a quantization step for quantizing neural network parameter data, and network configuration information including the par...

Full description

Saved in:
Bibliographic Details
Main Authors WANG MENGXION, MORIYA YOSHIMI, SUGIMOTO KAZUO, MINESAWA AKIRA
Format Patent
LanguageEnglish
Japanese
Published 08.04.2021
Subjects
BASIC ELECTRONIC CIRCUITRY
CALCULATING
CODE CONVERSION IN GENERAL
CODING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
DECODING
ELECTRICITY
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:To provide compression data which allows for highly compressing neural network parameter data.SOLUTION: Compressed data is generated by encoding quantization information defining a quantization step for quantizing neural network parameter data, and network configuration information including the parameter data quantized in the quantization step defined in the quantization information.SELECTED DRAWING: Figure 1 【課題】ニューラルネットワークのパラメータデータを高圧縮することができる圧縮データを提供する。【解決手段】圧縮データは、ニューラルネットワークのパラメータデータを量子化するときの量子化ステップが定義された量子化情報と量子化情報における量子化ステップで量子化されたパラメータデータを含むネットワーク構成情報とを符号化して生成される。【選択図】 図１
Bibliography:Application Number: JP20200206026