Schedule
| Saturday (July 05) | Sunday (July 06) | Monday (July 07) | Tuesday (July 08) | |
|---|---|---|---|---|
| 09:30 AM | Coffee | Coffee | Coffee | Coffee |
| 10:00 AM | Shashwat Agarwal Crypto Primitives - I [Slides] |
Vireshwar Kumar Security - I [Slides] |
Prateek Saxena On Provably Strong ML Security & Privacy Defenses [Talk details] |
Swaprava Nath Incentivize Contribution and Learn Parameters Too: Federated Learning with Strategic Data Owners [Talk details] |
| 11:15 AM | Coffee Break | Coffee Break | Coffee Break | Coffee Break |
| 11:45 AM | Anish Banerjee Crypto Primitives - II [Slides] |
Vireshwar Kumar Security - II [Slides] |
Reetika Khera Data and privacy:Putting Markets in (their) Place [Talk details, References] |
Arpita Patra Secure Multi-party Computation [Talk details] |
| 1:00 PM | Lunch Break | Lunch Break | Lunch Break + Group photograph (2 PM) | Lunch Break |
| 2:30 PM | Rohit Vaish Differential Privacy - I [Slides] |
Rohit Vaish Fairness in ML [Slides] |
Sikhar Patranabis Privacy-Preserving Search over Outsourced Encrypted Data [Talk details] |
Amit Kumar TBA |
| 3:45 PM | Coffee Break | Coffee Break | Coffee Break | Coffee Break |
| 4:15 PM | Amit Kumar Differential Privacy - II [Slides] |
Tarun Mangla Network Security [Slides] |
Kaustubh Beedkar TBA |
Vikash Chourasia TBA [Talk details] |
| 5:30 PM | Coffee Break | Coffee Break | Coffee Break | Coffee Break |
*The venue for all sessions is Room 501, Bharti Building.
Data and privacy: Putting Markets in (their) Place [Back to Schedule]
Reetika Khera (IIT Delhi)
Abstract: Should privacy be a tradeable right? It is believed that the economic opportunities presented by the rise of the digital technologies and of the digital economy on the one hand, and of data mining capabilities on the other, need to be harnessed. It is seen as a case of missing markets by some where appropriate markets with adequate rules and regulations should be put in place. In this paper, I argue that the creating market for personal data, amounts to creating a market for trading privacy. A market for personal data/ privacy has all the characteristics of what Debra Satz (2010) characterizes as "noxious markets". Following others including Bowles, Hausman and MacPherson and Sandel, I argue that the market for personal data should be included in the debates on moral limits to markets.
References:
Incentivize Contribution and Learn Parameters Too: Federated Learning with Strategic Data Owners [Back to Schedule]
Swaprava Nath (IIT Bombay)
Abstract: Classical federated learning (FL) assumes that the clients have a limited amount of noisy data with which they voluntarily participate and contribute towards learning a global, more accurate model in a principled manner. The learning happens in a distributed fashion without sharing the data with the center. However, these methods do not consider the incentive of an agent for participating and contributing to the process, given that data collection and running a distributed algorithm is costly for the clients. The question of rationality of contribution has been asked recently in the literature and some results exist that consider this problem. This paper addresses the question of simultaneous parameter learning and incentivizing contribution, which distinguishes it from the extant literature. Our first mechanism incentivizes each client to contribute to the FL process at a Nash equilibrium and simultaneously learns the model parameters. However, this equilibrium outcome can be away from the optimal, where clients contribute with their full data and the algorithm learns the optimal parameters. We propose a second mechanism with monetary transfers that is budget balanced and enables the full data contribution along with optimal parameter learning. Large scale experiments with real (federated) datasets (CIFAR-10, FeMNIST, and Twitter) show that these algorithms converge quite fast in practice, yield good welfare guarantees, and better model performance for all agents.
Joint work with Drashthi Doshi, Aditya Vema Reddy Kesari, Avishek Ghosh, and Suhas S Kowshik
Bio: Swaprava is an Associate Professor at the Department of Computer Science and Engineering, IIT Bombay. Before this, he was a faculty member at the Dept. of CSE, IIT Kanpur. Even earlier, he held postdoctoral positions at Carnegie Mellon University and Indian Statistical Institute, New Delhi, and finished his PhD from the Dept. of CSA, IISc Bangalore. His research interest lies at the intersection of economics and computation, which has several applications in social, industrial, and computational paradigms. Apart from academic positions, Swaprava also has experience in the industry. He has worked at Xerox Research Centre Europe and Cisco Systems India. He has been recipients of Fulbright-Nehru post doctoral grant, Tata Consultancy Services PhD Fellowship, and the Honorable Mention Award of Yahoo! Key Scientific Challenges Program.
Privacy-Preserving Search over Outsourced Encrypted Data [Back to Schedule]
Sikhar Patranabis (IBM Research)
Abstract: Consider a scenario where an organization outsources a relational database with private records to an untrusted cloud server for storage and processing. The organization wishes to leverage the server's computational capabilities to query the database, but while revealing as little information as possible about the data and the queries to the server. In this talk, I will cover searchable symmetric encryption (SSE) – a class of cryptographic solution that allows (symmetrically) encrypting a database while retaining the ability to query it in a privacy-preserving manner without decrypting it. SSE has been studied academically for nearly 20 years, and is now also beginning to see industry adoption, most notably as part of MongoDB's queryable encryption offering. The talk will explore various aspects of SSE that a designer must keep in mind, the most important (and challenging) being striking the right balance between functionality, efficiency, and security.
Secure Multi-party Computation [Back to Schedule]
Arpita Patra (Indian Institute of Science)
Abstract: Secure Multi-party Computation (MPC) is the standard-bearer and holy-grail problem in Cryptography that permits a collection of data-owners to compute a collaborative result, without any of them gaining any knowledge about the data provided by the other, except what is derivable from the result of the computation.
This talk will discuss the fundamental concept of garbled circuits and the Yao's two-party computation which is the first MPC construct in the literature.
Bio: Arpita Patra is currently a Professor of Computer Science at the Indian Institute of Science (IISc). She was a Visiting Faculty at Silence Laboratories, Singapore in 2024, and a Visiting Faculty Researcher at Google Research from 2022 to 2023.
Her research interests lie in cryptography, with a specific focus on both the theoretical foundations and practical applications of secure multiparty computation protocols. She earned her Ph.D. from the Indian Institute of Technology (IIT) Madras and subsequently held postdoctoral positions at leading institutions including the University of Bristol (UK), ETH Zurich (Switzerland), and Aarhus University (Denmark).Her contributions have been recognized through numerous honors, including: Prof. S. K. Chatterjee Award for Outstanding Woman Researcher/Industry Leader by IISc (2023); Google Privacy Research Faculty Award (2023); J.P. Morgan Chase Faculty Award (2022); SONY Faculty Innovation Award (2021); Google Research Award (2020); NASI Young Scientist Platinum Jubilee Award (2018); SERB Women Excellence Award (2016); INAE Young Engineer Award (2016). She is also affiliated with esteemed scientific bodies including the Indian Academy of Sciences (IAS), the Indian National Academy of Engineering (INAE), and The World Academy of Sciences (TWAS).
In addition to her research, she has coauthored two academic textbooks: Secure Multiparty Computation against Passive Adversaries (Springer, 2023) and Fault Tolerant Distributed Consensus in Synchronous Networks (Springer, 2025).
On Provably Strong ML Security & Privacy Defenses [Back to Schedule]
Prateek Saxena (National University of Singapore)
Abstract: ML systems are being rapidly deployed in security-sensitive applications, such as for authenticating human users. In these applications, ML systems are expected to safeguard user privacy and eliminate the risk of unauthorized backdoors. This talk focuses on two prominent attack classes---ML poisoning and model inversion attacks---that feature on the 'OWASP Top 10' threats to modern ML systems. Surprisingly, while these threats have been known for nearly a decade, no provably strong defenses for them have been shown yet. In this talk, I will highlight some of our recent conceptual and algorithmic results that can be utilized to build such defenses. In particular, I will present one of the first algorithms that generically defeats many ML poisoning attacks (including backdooring attacks) and has a practical runtime during training. I will also present a conceptual cryptographic primitive that can generically defeat model inversion, and outline remaining gaps in making it practical. The takeaway is that tractable algorithms in high-dimensional vector spaces are central to designing provable ML defenses.
Bio: Prateek Saxena is an Associate Professor in the Department of Computer Science at the National University of Singapore (NUS), where his research primarily focuses on computer security. His current research thrusts include designing algorithmic techniques for ML security/privacy, designing novel security processors, and automatic program translation for security. His past works have directly led to the founding of several research-backed startups and has been integrated into the commodity web platform developed by companies such as Google. Prateek has received numerous accolades for his past work, including the MIT TR35 Asia Award, the Google Security and Privacy Researcher Award, and the David J. Sakrison Award for doctoral research work at UC Berkeley. He has served on the program committees of security conferences, such as ACM CCS, IEEE S&P, and USENIX Security. Most recently, he served as an ACM CCS Area Chair for ML security in 2020 and 2021.
TBA [Back to Schedule]
Vikash Chourasia (Ministry of Electronics & Information Technology)
Bio: Vikash Chourasia is a Scientist at the Ministry of Electronics & Information Technology, Government of India. He holds a Master of Science degree in Cyber Laws & Information Security from the Indian Institute of Information Technology, Allahabad, India. In addition to his strong foundation in information security, he also holds an MBA degree, demonstrating his keen interest in management studies.
Prior to joining the government sector, Vikash gained experience as an Information Security auditor at leading IT companies. Currently, at the Ministry of Electronics and Information Technology (MeitY), he holds dual roles: one as Staff Officer to the Secretary and another as Scientist 'D'. In these capacities, Vikash is deeply involved in the development and formulation of the Digital Personal Data Protection Act 2023 and its associated rules. His contributions are instrumental in shaping and implementing legislation designed to safeguard digital personal data in India.