Thomas Chen

Title: The Innovative Contributions of Thomas Chen from Ann Arbor, MI

Introduction

Thomas Chen is a prominent inventor based in Ann Arbor, Michigan, known for his significant contributions to the field of memory systems and video inference processing. With three patents to his name, Chen has established himself as a leading figure in the realm of technological innovation. His work is primarily focused on enhancing computational efficiency, particularly in applications that require extensive signal processing and machine learning.

Latest Patents

Thomas Chen's latest patents demonstrate his cutting-edge contributions to memory technology and video processing. One of his notable inventions is the "SRAM-based process in memory system." This innovation addresses the computational and memory demands associated with numerous multiply-accumulate (MAC) operations, which are essential in various applications, including machine learning and scientific computing. By enabling direct computation on large data arrays, his process in memory technique significantly reduces the costly overhead associated with data movement.

To facilitate parallel multi-bit MAC operations, Chen's design utilizes width- and level-modulating memory word lines, along with a delay-locked loop to produce fine unit pulses for effective driving of memory word lines. Additionally, the invention incorporates a dual-ramp single-slope ADC for bit line output conversion. This concept has been prototyped in a 180 nm CMOS test chip featuring four compute-SRAMs capable of supporting 128 parallel 5 b × 5 b MAC operations, delivering a performance with a consumption of just 16.6 mW at 200 MHz.

Another innovative patent is his design for a "sparse video inference processor for action classification and motion tracking." This chip is engineered to efficiently extract spatio-temporal features from video data, enhancing action classification and motion tracking capabilities. The processor employs a recurrent neural network structured in three processing layers, where high sparsity is enforced to drastically reduce complexity. In practice, all multiply-accumulate functions are substituted with select-accumulates (SA), achieving remarkable efficiency. This invention was validated with a 3.98 mm² 40 nm CMOS chip, equipped with an Open-RISC processor for software-defined control and classification.

Career Highlights

Throughout his career, Thomas Chen has collaborated with prestigious organizations, including the University of Michigan and Intel Corporation, allowing him to develop and refine his technological innovations. His roles in these esteemed institutions have fostered a rich environment for research and development in memory systems and video processing technologies.

Collaborations

Thomas Chen has had the privilege of collaborating with notable colleagues in the field, including Zhengya Zhang and Gregory Kengho Chen. These partnerships have contributed to the advancement of his projects, leading to transformative innovations that impact the industry.

Conclusion

In conclusion, Thomas Chen's contributions to the fields of memory systems and video processing are noteworthy. His inventive spirit and commitment to advancing technology not only enhance computational processes but also pave the way for new applications in machine learning and video analysis. As an inventor, his work exemplifies the essence of innovation in today’s rapidly evolving technological landscape.