Past Member

Praneeth Vepakomma

Former Research Assistant
  • Camera Culture

Update: Please visit for most up-to-date details around my research.

 Bio: Praneeth Vepakomma is a PhD student at MIT and works on distributed and private computation. He has extensive industrial experience across Meta, Apple, Amazon, Motorola Solutions, Corning and several startups. His research focuses on developing algorithms for distributed computation in statistics & machine learning under constraints of privacy, & efficiency. He won the Meta PhD research fellowship in Applied Statistics, two SERC Scholarships (for Social and Ethical Responsibilities of Computing) from MIT's Schwarzman college of computing. He co-founded a research based non-profit (Integrity Distributed) that won the Financial Times Digital Innovation Award. He won a Best Student Paper Award at FL-IJCAI, a Baidu Best Paper Award at NeurIPS-SpicyFL and a Best Paper Runner Up Award at FG-2021. His technical work is inspired by foundations of non-asymptotic statistics, randomized algorithms, learning augmented algorithms, combinatorics, and at times just by systems design. He has organized several workshops at ICLR, ICML, IJCAI, CVPR and NeurIPS. 

  1. Posthoc privacy guarantees for collaborative inference with modified Propose-Test-Release, (PDF), @NeurIPS 2023 Conference, Thirty-seventh Conference on Neural Information Processing Systems, Abhishek Singh, Praneeth Vepakomma, Vivek Sharma, Ramesh Raskar -Topic: Differential privacy, formalizes privacy of informal ML pipelines, Distributed/Collaborative Inference (2023)
  2. Parallel quasi-concave set function optimization for scalability even without submodularity (PDF), @IEEE High Performance Extreme Computing Conference, Praneeth Vepakomma, Yulia Kempner, Rodmy Paredes Alfaro, Ramesh Raskar -Topic: Parallel Combinatorial Optimization (2023)
  3. Private independence testing across two parties (PDF), Praneeth Vepakomma, Mohammad Mohammadi Amiri, Clément L. Canonne, Ramesh Raskar, Alex Pentland - Topic: Statistics, Hypothesis Testing, Differential Privacy, Independence Testing (2022)
  4. The privacy-welfare trade-off: Effects of differential privacy on influence & welfare in social choice, Ibrahim Suat Evren,  Praneeth Vepakomma, Ramesh Raskar (2022)
  5. PrivateMail: Differentially private supervised manifold learning of deep features with privacy, @AAAI 2022, 36th AAAI Conference on Artificial Intelligence, (AAAI 2022) (PDF).
  6. LocFedMix-SL: Localize, Federate, and Mix for Improved Scalability, Convergence, and Latency in Split Learning@ WWW 2022 : International World Wide Web Conference/The Web Conf(WWW 2022), Seungeun Oh, Jihong Park, Praneeth Vepakomma, Sihun Baek, Ramesh Raskar, Mehdi Bennis and Seong-Lyun
  7. Fedml: A research library and benchmark for federated machine learning (Baidu Best Paper Award at NeurIPS-SpicyFL 2020)  (PDF)
  8. NoPeek-Infer: Preventing face reconstruction attacks in distributed inference after on-premise training, (IEEE FG 2021) (PDF)  (Mukh Best Paper Runner Up Award)
  9. DISCO: Dynamic and Invariant Sensitive Channel Obfuscation for deep neural networks  IEEE Computer Vison and Pattern Recogniton Conference (CVPR 2021) (PDF),
  10. Supervised Dimensionality Reduction via Distance Correlation MaximizationElectronic Journal of Statistics, volume 12 No.1, Pages 960--984, The Institute of Mathematical Statistics and the Bernoulli Society, 2018 (Journal) (PDF)   5.    Private measurement of nonlinear correlations between data hosted across multiple parties (PDF)      
  11. Advances and open problems in federated learning (PDF)( Foundations and Trends in Machine Learning -FnTML, 2020)
  12. AirMixML: Over-the-Air Data Mixup for Inherently Privacy-Preserving Edge Machine Learning(PDF),  IEEE Global Communications Conference (IEEE GLOBECOM 2021)
  13. Parallel Quasi-concave set optimization: A new frontier that scales without needing submodularity (PDF)(SubsetML @ ICML)
  14. DAMS: Meta-estimation of private sketch data structures for differentially private COVID-19 contact tracing(PRIML and PPML joint edition, NeurIPS-2020) (PDF)
  15. Splintering with distributions: A stochastic decoy scheme for private computation
  16. Assessing Disease Exposure Risk With Location Histories And Protecting Privacy: A Cryptographic Approach In Response To A Global Pandemic
  17. Split learning for health: Distributed deep learning without sharing raw patient data(PDF) (Project page) (ICLR Workshop)
  18. Combinatorics of Distance Covariance: Inclusion-Minimal Maximizers of Quasi-Concave Set Functions for Diverse Variable Selection ,  (Discrete Applied Mathematics, Journal)  (PDF)
  19. PPContactTracing: A privacy-preserving contact tracing protocol for covid-19 pandemic
  20. No Peek: A Survey of private distributed deep learning (PDF)
  21. Optimal bandwidth estimation for a fast manifold learning algorithm to detect circular structure in high-dimensional data (PDF)
  22. A Fast Algorithm for Manifold Learning by Posing it as a Symmetric Diagonally Dominant Linear System Applied and Computational Harmonic Analysis (PDF)
  23. A-Wristocracy: Deep Learning on Wrist-worn Sensing for Recognition of User Complex Activities (PDF)(IEEE Body Sensor Networks)
  24. Apps gone rogue: Maintaining personal privacy in an epidemic
  25. ExpertMatcher: Automating ML Model Selection for Clients using Hidden Representations, (2019 ICLR Workshop)
  26. "Scoring Practices for Remote Sensing of Land Mines", Mathematical Problems in Industry, MPI Workshop, Duke University.
  27. A Review of Homomorphic Encryption Libraries for Secure Computation (PDF)

Talks & Professional Service

-Penn State University talk on Differential Privacy for Measuring Nonlinear Correlations between Sensitive Data at Multiple Parties. Talk video: Link: or Link: as part of Young Achievers Symposium series hosted by the Center for Socially Responsible Artificial Intelligence.

-PriSecML, Privacy and Security in ML Interest Group (London & the world): Private Measurement of Nonlinear Correlations in a Distributed Setting Link:
- RIKEN, AIP-Japan Link:

-Rutgers Mathematics dept. talk on Combinatorics of Distance Covariance, in Experimental Mathematics Seminar.
-Carnegie Mellon University, talk on Distributed & Privacy Preserving Computations, at See Below the Skin Expeditions.
-IBM Thomas J Watson Research Center, talk on Split Learning: A new resource efficient alternative for distributed ML, Workshop on Federated Learning and Analytics.
-INFORMS Annual Meeting, under Covid-19, Service Science Track on privacy preserving ML for health care: Private contact tracing and split learning.
-Tutorial at CVPR 2019 on distributed ML with federated learning and split learning.
-On Split Learning, a general talk at
Reviewer: Journal of Machine Learning Research (JMLR), AISTATS conference, IJCAI conference, Transactions on Dependable and Secure Computing, IEEE Internet of Things Journal, IEEE Intelligent Systems, IEEE Transactions on Services Computing
Organized: ICML International Workshop on Federated Learning for User Privacy and Data Confidentiality, 2021 Keynote Speakers: Sebastian U. Stich, Nic Lane, Ramesh Raskar, Ameet Talwalkar, Filip Hanzely, Dimitris Papailiopoulos and Salman Avestimehr
Organized: ICLR Workshop on Distributed and Private Machine Learning (DPML, 2021).  Keynote Speakers: David Evans, Lalitha Sankar, Gauri Joshi & Graham Cormode.
CVPR Tutorial On Distributed Private Machine Learning for Computer Vision: Federated Learning, Split Learning and Beyond, 2019. Tutorial Speakers: Brendan McMahan, Jakub Konečný, Ramesh Raskar, Praneeth Vepakomma, Otkrist Gupta, Hassan Takabi
Organized: Workshop on Split Learning for Distributed Machine Learning (SLDML’21). Keynote Speakers: Peter Kairouz, OpenMined, Geeta Chauhan and Supriyo Chakraborty.​

Sampling of problems: A small sampling of problems that I work on includes a.) Private independence testing and private k-sample testing in statistics, b.) Bridging privacy with social choice theory, c.) Private mechanisms for training and inference in ML, d.) Privately estimating non-linear measures of statistical dependence between multiple parties and e.) Split learning.

Visit for most up-to-date details.