Ming-Yu Liu

Title: Innovations and Contributions of Ming-Yu Liu

Introduction

Ming-Yu Liu, a prominent inventor located in Sunnyvale, CA, has made significant strides in the field of technology with an impressive portfolio of 41 patents. His inventive work primarily focuses on machine learning and video processing, contributing to advancements in various applications within these domains.

Latest Patents

Among his notable patents, Liu has developed a machine-learning-based architecture search method for neural networks. This innovative approach merges differentiable neural architecture search with reinforcement learning to identify network architectures that excel in both accuracy and latency. Furthermore, by utilizing an objective function tied to generalization error, Liu's method effectively avoids architectures that may lead to overfitting. Another significant patent involves frame selection for streaming applications. This invention addresses the challenge of identifying blurred frames in video sequences by leveraging a motion variance threshold that dynamically adapts based on a moving average of variance of motion determined from reference frames.

Career Highlights

Ming-Yu Liu has contributed his expertise to several influential companies, including Nvidia Corporation and Mitsubishi Electric Research Laboratories, Inc. Throughout his career, he has actively participated in the evolution of technologies that enhance machine learning and streaming video processing.

Collaborations

Liu has collaborated with notable colleagues such as Jan Kautz and Oncel C Tuzel, enhancing the collective knowledge and innovation within their projects and furthering advancements in their fields.

Conclusion

With a robust record of patents and collaborations, Ming-Yu Liu continues to be a driving force in the technology landscape. His innovative contributions not only demonstrate his expertise but also pave the way for future advancements in machine learning and video processing applications.