Xinxuan Wu

Title: Xinxuan Wu: Inventor in Distributed and Neural Network Training

Introduction

Xinxuan Wu is an inventor based in Beijing, China, whose patent work focuses on artificial intelligence, deep learning, and distributed model training. His portfolio includes 2 patents addressing methods and systems for training machine-learning models.

Latest Patents

Wu’s listed patents include “Apparatus and Method for Distributed Model Training, Device, and Computer Readable Storage Medium.” The invention describes a distributed reader, trainer, and parameter server that operate independently. The reader supplies training samples to trainers, which generate gradient information; the parameter server uses that information to update the model’s parameters.

His other listed patent is “Neural Network Training Method and Apparatus, Electronic Device, Medium and Program Product.” It describes using a deep reinforcement learning model to select processing units for neural-network layers. The selection is based on training duration and the cost of different types of processing units.

Career Highlights

Wu is associated with Beijing Baidu Netcom Science Technology Co., Ltd., also known as Beijing Baidu Netcom Science and Technology Co., Ltd. His patent work reflects technical interests in improving the organization and efficiency of model training.

Collaborations

Dianhai Yu and Xuefeng Yao are identified as Wu’s coworkers. Their shared professional context connects them to the broader work surrounding artificial intelligence and machine-learning technologies.

Conclusion

Xinxuan Wu’s patents address distributed training architecture and the selection of processing resources for neural networks, highlighting his contributions to practical approaches in artificial intelligence.