2431 EECS 1301 Beal Avenue Ann Arbor, MI 48109-2122 [email protected]
Research Interest:
Efficient AI accelerator design, low power VLSI system design
Current Research:
Efficient Neural Network Accelerator via Software-Hardware Co-design
In recent times, neural networks have become increasingly computation- and memory-intensive. This project focuses on designing efficient neural network accelerators, encompassing system design (audio, image, natural language, and multimodal processing systems), algorithm design (model quantization, sparsity), architecture design (parallel computing architecture, sparse computation architecture), and chip design (low-power circuit techniques).