Pratik Chaudhari

I am a PhD candidate in the Computer Science department at UCLA and I work with prof. Stefano Soatto in the Vision Lab on deep learning.

I have an Engineer and a Master's degree in Aeronautics & Astronautics from MIT where I worked with prof. Emilio Frazzoli at the Laboratory of Information and Decision Systems (LIDS). I was in the Aerospace Engineering department at IIT Bombay for my undergraduate studies until 2010.

Over the past few years, I have worked on self-driving cars in the areas of computer vision, motion-planning algorithms using temporal logic, and stochastic estimation at nuTonomy Inc. and SMART, Singapore. Some of my favorite techniques to tackle research problems draw from statistical physics and optimization, and in general, topics in applied probability such as random matrix theory, large deviations and stochastic geometry excite me.

Resume

Contact

  pratikac at ucla dot edu

Web

GitHub Twitter Facebook Google+

Publications (Google Scholar)

Parle: parallelizing stochastic gradient descent
P. Chaudhari, C. Baldassi, R. Zecchina, S. Soatto, A. Talwalkar
arXiv:1707.00424
URL
Deep Relaxation: partial differential equations for optimizing deep neural networks
P. Chaudhari, A. Oberman, S. Osher, S. Soatto, G. Carlier
arXiv:1704.04932 (in submission)
URL

Short version, Principled Approaches to Deep Learning, ICML 2017
PDF
Entropy-SGD: Biasing gradient descent into wide valleys
P. Chaudhari, A. Choromanska, S. Soatto, Y. LeCun, C. Baldassi, C. Borgs, J. Chayes, L. Sagun, R. Zecchina
International Conference of Learning and Representations, 2017
URL Code
On the energy landscape of deep networks
P. Chaudhari, S. Soatto
arXiv:1511.06485
Advances in non-convex analysis and optimization, ICML 2016
URL
Incremental synthesis of minimum-violation control strategies for robots interacting with external agents
P. Chaudhari, T. Wongpiromsarn, E. Frazzoli
American Control Conference, 2014
PDF Code
Sampling-based algorithms for optimal motion planning using process algebra specifications
P. Chaudhari, V. Varricchio, E. Frazzoli
IEEE Conference on Robotics and Automation, 2014
PDF Video
Game theoretic controller synthesis for multi-robot motion planning
Part I : Trajectory based algorithms

M. Zhu, M. Otte, P. Chaudhari, E. Frazzoli
IEEE Conference on Robotics and Automation, 2014
URL
Incremental sampling-based algorithm for minimum-violation motion planning
L. Reyes-Castro, P. Chaudhari, J. Tumova, S. Karaman, E. Frazzoli, D. Rus
IEEE Conference on Decision and Control, 2013
URL Code Video
Sampling-based algorithms for continuous-time POMDPs
P. Chaudhari, S. Karaman, D. Hsu, E. Frazzoli
American Control Conference, 2013
PDF Code
Sampling-based algorithm for filtering using Markov chain approximations
P. Chaudhari, S. Karaman, E. Frazzoli
IEEE Conference on Decision and Control, 2012
PDF Code

Thesis

Algorithms for autonomous urban navigation with formal specifications
P. Chaudhari
Engineers's thesis, Aeronautics and Astronautics, MIT, 2014
PDF
Incremental sampling based algorithms for state estimation
P. Chaudhari
Master's thesis, Aeronautics and Astronautics, MIT, 2012
PDF

Talks and Posters

PDF Entropy-SGD: Biasing gradient descent into wide valleys
International Conference of Learning and Representations, 2017
PDF Visual representations: Defining properties and Deep approximations
International Conference of Learning and Representations, 2016
PDF Energy landscapes of deep networks
Special topics in Machine Learning seminar, UCLA
PDF Sampling-based algorithms: Planning for stochastic sytems and complex specifications
Automatic Control Group, KTH
PDF Approximate POMDP homomorphisms
Joint Inference and Control seminar, LIDS, MIT