-
公开(公告)号:US12299565B2
公开(公告)日:2025-05-13
申请号:US17135793
申请日:2020-12-28
Applicant: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTE
Inventor: Chao-Hung Chen , Ming-Chun Hsyu , Chien-Chih Huang , Wen-Pin Hsu , Chun-Te Yu
Abstract: A data processing method used in neural network computing is provided. During a training phase of a neural network model, a feedforward procedure based on a calibration data is performed to obtain distribution information of a feedforward result for at least one layer of the neural network model. During the training phase of the neural network model, a bit upper bound of a partial sum is generated based on the distribution information of the feedforward result. During an inference phase of the neural network model, a bit-number reducing process is performed on an original operation result of an input data and a weight for the neural network model according to the bit upper bound of the partial sum to obtain an adjusted operation result.
-
公开(公告)号:US10834202B2
公开(公告)日:2020-11-10
申请号:US16231449
申请日:2018-12-22
Applicant: Industrial Technology Research Institute
Inventor: Wei-Cheng Wang , Wen-Pin Hsu
IPC: G06F15/173 , H04L29/08 , H04W4/60 , H04L29/12 , H04W8/18
Abstract: A network service method includes: a core cloud server returns all running service identification information; a multi-access edge computing cloud server receives a service request from a terminal device; and the multi-access edge computing cloud server determines whether an application that corresponds to the service request exists in the multi-access edge computing cloud server, according to the service identification information reported by the core cloud server; after the multi-access edge computing cloud server integrates the current state with the application being run by the multi-access edge computing cloud server, the terminal device is directed to the multi-access edge computing cloud server, and the multi-access edge computing cloud server provides the application to the terminal device.
-