Illustrated portrait of Karthik

I'm Karthik, a PhD student at CVSSP, University of Surrey, UK.

I mostly think about the following questions:

  1. What properties of real data do neural networks encode that allow them to generalize?
  2. How do neural networks encode these properties?
  3. Is it possible to create more efficient training data for neural networks synthetically?

I occasionally work on other ideas as well.

  • World Models
  • Generative Modelling

This is just a space for me to communicate my thoughts on these topics, mostly because sometimes my friends and girlfriend get bored of hearing me talk about this.