This inventor holds 2 USPTO granted patents. Top assignee: National Tsing Hua University. Active years: 2022-2023.
Location History:
- Taoyuan, TW (2022)
- Hsinchu, TW (2023)
Company Filing History:
Years Active: 2022-2023
Title: Innovations by Wei-Hsing Huang
Introduction
Wei-Hsing Huang is a notable inventor based in Taoyuan, Taiwan. He has made significant contributions to the field of computing-in-memory neural networks. With a total of 2 patents, his work focuses on enhancing the efficiency and functionality of neural network systems.
Latest Patents
Wei-Hsing Huang's latest patents include a dynamic gradient calibration method for computing-in-memory neural networks. This method updates a plurality of weights in a computing-in-memory circuit based on inputs corresponding to a correct answer. The process involves a forward operating step that performs a bitwise multiply-accumulate operation on divided inputs and weights, generating clamped multiply-accumulate values. These values are then compared to the correct answer to produce loss values. The backward operating step calculates a weight-based gradient to update the weights accordingly.
Another significant patent is related to a memory unit designed for multi-bit convolutional neural network applications. This memory unit is controlled by two word lines and includes a memory cell that stores a weight. The transpose cell connected to the memory cell receives the weight and generates a multi-bit output value based on the input value and the weight.
Career Highlights
Wei-Hsing Huang is currently affiliated with Tsinghua University, where he continues to advance research in computing technologies. His innovative approaches have positioned him as a key figure in the development of neural network systems.
Collaborations
He has collaborated with notable colleagues, including Meng-Fan Chang and Yung-Ning Tu, contributing to various research projects and innovations in the field.
Conclusion
Wei-Hsing Huang's contributions to computing-in-memory neural networks demonstrate his commitment to advancing technology. His patents reflect a deep understanding of neural network systems and their applications.