Diganta Misra

Neural network loss landscape visualisations from my first paper (Mish, 2020)

In mathematics the art of proposing a question must be held of higher value than solving it.    –Georg Cantor

I’m extremely fascinated and curious of understanding the underlying theoretical mechanisms governing deep learning methods especially in the domain of Mean Field Theory, Non Convex Optimization and Non-Linear Dynamics.

I broadly work on theoretical and analytical deep learning with focus on but not limited to the following domains:

• Non-linear dynamics & Loss Landscapes
• Adversarial Robustness
• Sparsity

Specifically, my current work is centred around using Measure Theory to prove theoretical robustness guarantees in neural networks.

I also like to read a lot of fiction and compose abstract artworks. I am also a trained pianist and prefer Lo-Fi/ Phonk music.