Yuhao Sun

Ph.D Student

The brain exhibits activity and structural features across various scales, from macroscopic to microscopic, that are not entirely equivalent to current artificial intelligence, and demonstrates strong integrative intelligence capabilities. My research begins at the microscopic scale, where I have developed a brain-inspired artificial intelligence platform based on micro-scale encoding rules, addressing fundamental tool issues in heterogeneous hardware-software hybrid structures. At the mesoscopic scale, I proposed brain-inspired learning algorithms based on sampling and gradient estimation, establishing a brain-inspired sampling learning theory that overcomes the conflict of the infeasibility of backpropagation in brain-like learning frameworks. At the macroscopic scale, I introduced a stochastic dynamical description method based on non-stationary, non-Gaussian stochastic processes for macroscopic brain activities, such as EEG signals, providing a theoretical tool for further interpreting the frequency domain and oscillatory characteristics of brain activity.