While the current wave of AI has demonstrated the engineering power of brute-force scaling, it has also fallen into a "black box" dilemma—heavily reliant on massive computational resources and data, yet fundamentally lacking interpretability. My research is dedicated to returning to the first principles of physics and computational neuroscience. By drawing inspiration from the highly efficient mechanisms of the biological visual cortex, I explore the foundational architecture for next-generation machine vision. I firmly believe that the true path to Artificial General Intelligence (AGI) lies not merely in stacking parameters, but in revealing and predicting the dynamic evolution of the complex physical world through a few elegant mathematical equations.

To this end, my work focuses on breaking the barrier between pure analytical mathematics and the complex mechanisms of the biological brain. I aim to build brain-inspired visual predictive models that are highly interpretable, low-power, and compute-friendly. By injecting rigorous physical intuition into machine vision, I aspire to develop foundation models capable of seamless deployment on edge computing devices and robotic hardware, enabling machines to understand the world with the same minimalist yet profound mathematical elegance as a biological brain.