Hao Dong

Hao Dong is a Researcher affiliated with the University of California, Santa Barbara (UCSB) and Meta. His primary role involves advancing research in machine learning and GPU-accelerated computing. He is part of the Department of Statistics and Applied Probability at UCSB.

His research focuses on optimizing deep learning models through techniques like sparsity, attention mechanisms, and hardware acceleration. Key areas include transformer models, spiking neural networks, and graph neural networks. He has contributed to frameworks like H2Learn and fuseGNN, emphasizing efficiency in training and inference phases.

His publications consistently explore computational efficiency, scalability, and hardware-software co-design, particularly leveraging GPUs to accelerate neural network operations. Notable work includes dynamic sparse attention mechanisms and structured sparsity strategies to reduce computational costs while maintaining accuracy.

No scientific awards or grants are explicitly mentioned in the provided information. His academic contributions are centered on algorithmic innovation and practical implementation in high-performance computing environments.