This inventor holds 1 USPTO granted patent. Top assignee: National Tsing Hua University. Active years: 2022.
Company Filing History:
Years Active: 2022
Title: Innovations of Jing-Hong Wang in Neural Network Quantization
Introduction
Jing-Hong Wang is a notable inventor based in Hsinchu, Taiwan. He has made significant contributions to the field of neural networks, particularly in the area of quantization methods. His work focuses on enhancing the efficiency and accuracy of convolutional neural networks through innovative techniques.
Latest Patents
Wang holds a patent titled "Quantization method for partial sums of convolution neural network based on computing-in-memory hardware and system thereof." This patent introduces a quantization method that involves a probability-based quantizing step and a margin-based quantizing step. The probability-based quantizing step includes several stages such as network training, quantization-level generation, and accuracy generation. The margin-based quantizing step enhances the quantization process by adjusting quantization edges to improve accuracy. Notably, the second accuracy value generated from this method surpasses the first accuracy value, showcasing the effectiveness of his approach.
Career Highlights
Jing-Hong Wang is affiliated with Tsinghua University, where he continues to engage in research and development in the field of computing and neural networks. His academic background and ongoing projects contribute to advancements in technology and innovation.
Collaborations
Wang has collaborated with notable colleagues such as Meng-Fan Chang and Ta-Wei Liu. These partnerships enhance the research output and foster a collaborative environment for innovation.
Conclusion
Jing-Hong Wang's contributions to the field of neural networks through his patented quantization methods demonstrate his commitment to advancing technology. His work not only improves the efficiency of convolutional neural networks but also sets a foundation for future innovations in computing.