Yakun Shao

Title: Innovations by Yakun Shao in Neural Network Acceleration

Introduction

Yakun Shao is a prominent inventor based in Santa Clara, CA, known for his significant contributions to the field of neural network acceleration. With a total of four patents to his name, he has been instrumental in advancing the efficiency and performance of deep learning systems. His work primarily focuses on optimizing dataflows in distributed deep neural networks.

Latest Patents

One of Yakun Shao's latest patents is titled "Efficient Neural Network Accelerator Dataflows." This invention describes a distributed deep neural network (DNN) that utilizes a tile-based architecture comprising multiple chips. Each chip features a central processing element, a global memory buffer, and several additional processing elements. These additional processing elements are equipped with weight buffers, activation buffers, and vector multiply-accumulate units, allowing them to combine weight and activation values in parallel using stationary data flows.

Another notable patent is "Deep Neural Network Accelerator with Fine-Grained Parallelism Discovery." This patent outlines a sparse convolutional neural network accelerator system that dynamically identifies fine-grained parallelism in sparse convolution operations. The system efficiently determines matching pairs of non-zero input activations and weights from compacted input activation and weight arrays. It employs a scalable, dynamic parallelism discovery unit (PDU) that performs a parallel search to identify reducible input activation and weight pairs.

Career Highlights

Yakun Shao is currently employed at Nvidia Corporation, a leading company in graphics processing and AI technology. His work at Nvidia has allowed him to push the boundaries of what is possible in neural network acceleration, contributing to the company's reputation as a pioneer in the field.

Collaborations

Throughout his career, Yakun has collaborated with notable colleagues, including Joel S. Emer and Stephen William Keckler. These collaborations have further enriched his research and development efforts, leading to innovative solutions in deep learning technologies.

Conclusion

Yakun Shao's contributions to neural network acceleration exemplify the impact of innovative thinking in technology. His patents reflect a commitment to enhancing the efficiency of deep learning systems, positioning him as a key figure in the advancement of artificial intelligence.