Ya-Tse Wu

Title: Ya-Tse Wu: Advancing Machine-Learning Model Training

Introduction

Ya-Tse Wu is an inventor associated with Hsinchu, Taiwan, and Tsinghua University. His patent work addresses methods and tools for training machine-learning models.

Latest Patents

Wu is credited with 1 patent: “Method, Device and Non-Transitory Computer Readable Medium for Training Machine Learning Models.” The invention describes drawing samples from a first training set according to multiple weights during an initial training epoch. The set contains levels corresponding to those weights. After the epoch, the model is evaluated on a validation set to measure performance at each level. Those results are then used to update the weights, which guide sample selection in a subsequent epoch. The patent also describes an electronic device and a non-transitory computer-readable medium for implementing the method.

Career Highlights

The patent reflects an approach to machine-learning training that uses validation performance to adjust how training samples are selected across different levels. Wu is associated with Tsinghua University.

Collaborations

Wu’s listed coworker is Chi-Chun Lee. Their shared association connects their work to the research and innovation environment at Tsinghua University.

Conclusion

Ya-Tse Wu’s patent presents a performance-guided method for selecting training samples, alongside device and software-medium implementations, contributing to innovation in machine-learning model training.