About
Hello there! I’m Mark. I'm an undergraduate at UC Berkeley studying math, computer science, and physics. My interests lie at the intersection of all of these fields. I’m fascinated by emergent phenomena that arise from complex (usually nonlinear) systems. I’m particularly interested in the learning capabilities of neural networks and more broadly, the mysteries of deep learning.
The empirical success of deep learning is undeniable. But like many engineering successes of the past, our theoretical understanding lags behind. A genuine scientific theory of deep learning would be paradigm-altering. It would drive significant engineering advances and provide us with a window into the very nature of intelligence. Such a theory would be layered, the way the natural sciences are: psychology, biology, and physics each building a map at a different resolution. I’m most interested in the bottom layer resembling a first-principles physics of learning, an emerging discipline now called learning mechanics which treats deep learning the way physics treats the natural world; seeking compact mathematical principles, tight connections between theory and experiment, and simple, intuitive explanations for complex phenomena.
I’m currently working on learning mechanics at Feature Lab (FLAB), housed within the Redwood Center for Theoretical Neuroscience.
