METHOD AND DEVICE FOR QUANTIZING PARAMETER OF NEURAL NETWORK

To provide a method and a device for quantizing the parameter of a neutral network.SOLUTION: There are provided a method and a neural network device for quantizing the updated parameter in a learning process or an estimating process by using the bit shift value of the parameter, updating a fixed-poi...

Full description

Saved in:
Bibliographic Details
Main Authors PARK HYUNSUN, KANG SHINHAENG, RI TOSHIYUKI
Format Patent
LanguageEnglish
Japanese
Published 23.04.2020
Subjects
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
ELECTRIC DIGITAL DATA PROCESSING
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:To provide a method and a device for quantizing the parameter of a neutral network.SOLUTION: There are provided a method and a neural network device for quantizing the updated parameter in a learning process or an estimating process by using the bit shift value of the parameter, updating a fixed-point system, and by the updated fixed-point system.SELECTED DRAWING: Figure 5 【課題】 ニューラルネットワークのパラメータを量子化する方法及びその装置を提供する。【解決手段】 パラメータのビットシフト値を利用し、固定小数点形式をアップデートし、アップデートされた固定小数点形式により、学習過程または推論過程でアップデートされたパラメータを量子化する方法、及びそのためニュ−ロルネットワーク装置である。【選択図】 図５
Bibliography:Application Number: JP20190189522