Neuromorphic Computing / Computer Hardware / Sparse Coding / Neural Networks as Dynamical Systems

My research explores how brain-like information processing can be introduced into AI through the co-design of algorithms and hardware. Modern AI research has been shaped by GPUs, which favor massively parallel computation. As a result, principles such as sparsity, iterative and recurrent computation, and the close integration of computation and memory have received less attention. This contributes to unresolved problems including rising energy consumption and the difficulty of long-term continual learning. I study brain-inspired algorithms such as Spiking Neural Networks and Predictive Coding Networks, and accelerate them on emerging AI hardware designed beyond the conventional GPU paradigm. My goal is to narrow the gap between biological brains and practical AI.

• Lab website