Projects

XNet : Efficient Deep Networks

Use Expander graphs for making Deep Neural Networks Efficient

Deep Reinforcement Learning

Surveying Deep learning methods used in Reinforcement Learning

Model Compression

Make deep learning models deployable in constrained memory devices.

Hardness of Approximate Coloring

Improved lowerbounds for graph/hypergraph coloring.

Physarum Computer

Formal proof for slime-molds finding shortest paths in maze.

Probabilitically Checkable Proofs

Improved PCPs using low-degree codes and using product constructions.

Streaming Algorithms

Few pass, small memory algorithms for big data.

Selected Publications

All Publications

Super-polylogarithmic hypergraph coloring hardness via low-degree long codes
SIAM Journal on Computing. 2017.

Preprint PDF Project Slides

Hardness of Approximate Coloring
Phd Thesis at Tata Institute of Fundamental Research, Mumbai. 2016.

PDF Project Project Poster Slides

Physarum Can Compute Shortest Paths
Journal of Theoretical Biology, volume 309-0, pages 121 - 133, 2012,. 2012.

PDF Project

Talks

Deep Learning Programming & Deployment
Sep 1, 2017
Introduction to Machine Learning
Jul 9, 2017
Model Compression
Jul 15, 2017

Contact