Invention Grant
- Patent Title: Techniques for parameter set and header design for compressed neural network representation
-
Application No.: US17324623Application Date: 2021-05-19
-
Publication No.: US11611355B2Publication Date: 2023-03-21
- Inventor: Byeongdoo Choi , Wei Wang , Wei Jiang , Stephan Wenger , Shan Liu
- Applicant: TENCENT AMERICA LLC
- Applicant Address: US CA Palo Alto
- Assignee: TENCENT AMERICA LLC
- Current Assignee: TENCENT AMERICA LLC
- Current Assignee Address: US CA Palo Alto
- Agency: Sughrue Mion, PLLC
- Main IPC: H03M7/00
- IPC: H03M7/00 ; H03M7/30 ; H04L65/75

Abstract:
Systems and methods for encoding and decoding neural network data is provided. A method includes: receiving a neural network representation (NNR) bitstream including a group of NNR units (GON) that represents an independent neural network with a topology, the GON including an NNR model parameter set unit, an NNR layer parameter set unit, an NNR topology unit, an NNR quantization unit, and an NNR compressed data unit; and reconstructing the independent neural network with the topology by decoding the GON.
Public/Granted literature
- US20210399739A1 TECHNIQUES FOR PARAMETER SET AND HEADER DESIGN FOR COMPRESSED NEURAL NETWORK REPRESENTATION Public/Granted day:2021-12-23
Information query