Quan Lin

Title: Innovations by Quan Lin in Efficient Deconvolution Implementation

Introduction

Quan Lin is an accomplished inventor based in San Jose, CA. He has made significant contributions to the field of hardware acceleration through his innovative patent. His work focuses on methods and devices that enhance the efficiency of general deconvolution implementations.

Latest Patents

Quan Lin holds a patent titled "Methods and devices for efficient general deconvolution implementation on hardware accelerator." This patent describes a method for implementing a devolution operation to obtain a multi-dimensional output feature map. The process involves subsampling a multi-dimensional kernel into non-overlapping multi-dimensional sub-kernels. By applying a stride-dependent virtual padded devolution operation, the method efficiently generates multi-dimensional sub-output feature maps. Additionally, it utilizes a predefined multi-dimensional interleave-concatenation method to reorder these sub-output feature maps into the final output feature map.

Career Highlights

Quan Lin is associated with Beijing Transtreams Technology Co. Ltd., where he continues to develop innovative solutions in hardware acceleration. His expertise in deconvolution methods has positioned him as a key player in the industry. He has successfully contributed to advancing technology in this area.

Collaborations

Quan Lin has worked alongside talented colleagues such as Shiya Liu and Ming Kai Hsu. Their collaborative efforts have further enhanced the development of efficient hardware solutions.

Conclusion

In summary, Quan Lin's innovative work in efficient deconvolution implementation has made a significant impact in the field of hardware acceleration. His contributions continue to shape advancements in technology.