Chirag Pabbaraju Email: cpabbara [at] stanford [dot] edu I'm a fifth year PhD student at Stanford University in the CS Theory group. I feel very lucky to be co-advised by Moses Charikar and Gregory Valiant. My primary research interests are in statistical and algorithmic aspects of learning theory. More broadly, I like thinking about various topics in theoretical computer science and machine learning. My research is generously supported by a Google PhD Fellowship.

Research

• Agnostic Language Identification and Generation
  Mikael Møller Høgsgaard, Chirag Pabbaraju
  Preprint
• A Characterization of List Language Identification in the Limit
  Moses Charikar, Chirag Pabbaraju, Ambuj Tewari
  Preprint
• Learning with Monotone Adversarial Corruptions
  Kasper Green Larsen, Chirag Pabbaraju, Abhishek Shetty
  ALT 2026
• Pareto-optimal Non-uniform Language Generation
  Moses Charikar, Chirag Pabbaraju
  ALT 2026
• A Simple Geometric Proof of the Optimality of the Sequential Probability Ratio Test for Symmetric Bernoulli Hypotheses
  Chirag Pabbaraju, Gregory Valiant, Rishi Verma
  SOSA 2026
• Testing with Non-identically Distributed Samples
  Shivam Garg, Chirag Pabbaraju, Kirankumar Shiragur, Gregory Valiant
  TMLR 2025 (J2C Certification)
• Lower Bounds for Greedy Teaching Set Constructions
  Spencer Compton, Chirag Pabbaraju, Nikita Zhivotovskiy
  COLT 2025
• Proofs as Explanations: Short Certificates for Reliable Predictions
  Avrim Blum, Steve Hanneke, Chirag Pabbaraju, Donya Saless
  COLT 2025
• Exploring Facets of Language Generation in the Limit
  Moses Charikar, Chirag Pabbaraju
  COLT 2025
• Relating Misfit to Gain in Weak-to-Strong Generalization Beyond the Squared Loss
  Abhijeet Mulgund, Chirag Pabbaraju
  ICML 2025
• New and Improved Bounds for Markov Paging
  Chirag Pabbaraju, Ali Vakilian
  ICALP 2025
• A Characterization of List Regression
  Chirag Pabbaraju, Sahasrajit Sarmasarkar
  ALT 2025
• Embedding Probability Distributions into Low Dimensional \(\ell_1\): Tree Ising Models via Truncated Metrics
  Moses Charikar, Spencer Compton, Chirag Pabbaraju
  SODA 2025
• Quantifying the Gain in Weak-to-Strong Generalization
  Moses Charikar, Chirag Pabbaraju, Kirankumar Shiragur
  NeurIPS 2024
• Credit Attribution and Stable Compression
  Roi Livni, Shay Moran, Kobbi Nissim, Chirag Pabbaraju
  NeurIPS 2024
• Multiclass Learnability Does Not Imply Sample Compression
  Chirag Pabbaraju
  ALT 2024 (Outstanding Paper Award)
• Provable Benefits of Score Matching
  Chirag Pabbaraju, Dhruv Rohatgi, Anish Sevekari, Holden Lee, Ankur Moitra, Andrej Risteski
  SPIGM workshop, ICML 2023 (Oral)
  NeurIPS 2023 (Spotlight)
  
• Harnessing the Power of Choices in Decision Tree Learning
  Guy Blanc, Jane Lange, Chirag Pabbaraju, Colin Sullivan, Li-Yang Tan, Mo Tiwari
  NeurIPS 2023
• A Characterization of List Learnability
  Moses Charikar, Chirag Pabbaraju
  STOC 2023
• Pitfalls of Gaussians as a Noise Distribution in NCE
  Holden Lee, Chirag Pabbaraju, Anish Sevekari, Andrej Risteski
  SSL workshop, NeurIPS 2022
  ICLR 2023
• Universal Approximation for Log-concave Distributions using Well-conditioned Normalizing Flows
  Holden Lee, Chirag Pabbaraju, Anish Sevekari, Andrej Risteski
  INNF+ workshop, ICML 2021 (Spotlight)
  NeurIPS 2021
• Estimating Lipschitz Constants of Monotone Deep Equilibrium Models
  Chirag Pabbaraju, Ezra Winston, Zico Kolter
  ICLR 2021
• Efficient Semidefinite-programming-based Inference for Binary and Multi-class MRFs
  Chirag Pabbaraju, Po-Wei Wang, Zico Kolter
  NeurIPS 2020 (Spotlight)
• Learning Functions over Sets via Permutation Adversarial Networks
  Chirag Pabbaraju, Prateek Jain
  Preprint
• GesturePod: Enabling On-device Gesture-based Interaction for White Cane Users
  Shishir G. Patil, Don Kurian Dennis, Chirag Pabbaraju, Nadeem Shaheer, Harsha Vardhan Simhadri, Vivek Seshadri, Manik Varma, Prateek Jain
  UIST 2019
• Multiple Instance Learning for Efficient Sequential Data Classification on Resource-constrained Devices
  Don Kurian Dennis, Chirag Pabbaraju, Harsha Vardhan Simhadri, Prateek Jain
  NeurIPS 2018

Patents

• Image Segmention via Efficient Semidefinate-programming Based Inference for Binary and Multi-class Markov Random Fields
  Devin Wilmott, Chirag Pabbaraju, Po-Wei Wang, Zico Kolter
  United States Patent number 11587237 assigned to Robert Bosch GmbH

Otherwise....

In Summer 2025, I was a student researcher at Google Research, working with Pranjal Awasthi. In Summer 2024, I had the absolute pleasure of interning with Avrim Blum at TTIC. Before starting my PhD, I was a masters student in the Machine Learning Department at Carnegie Mellon University. I feel fortunate to have been mentored by Zico Kolter and Andrej Risteski there. Even before that, I was a Research Fellow amongst wonderful peers at Microsoft Research India. I am grateful to have been advised by Prateek Jain there. Lastly, I completed my undergraduate studies at BITS Pilani Goa Campus, which remains my favorite home till date.