sayan.biswas@epfl.ch or bizwas05@gmail.com

EPFL IC IINFCOM SACS,
BC 166 (Bâtiment BC), Station 14,
1015 Lausanne, Switzerland.

About Me

I am a Postdoctoral Researcher under the supervision of Anne-Marie Kermarrec at the SaCS lab at EPFL, Switzerland. Recently, I completed my PhD in Computer Science under the supervision of Catuscia Palamidessi from INRIA and École Polytechnique in France. Previously, I have been a Visiting Scholar at the School of Computing at Macquarie University in Sydney, Australia, collaborating with Annabelle McIver and Natasha Fernandes, and at WMG, the University of Warwick in Coventry, England, working with Carsten Maple and Graham Cormode.

My work primarily involves designing privacy-preserving techniques for analysing data and training models, with a focus on investigating and enhancing the trade-off between privacy and utility from a foundational perspective. In particular, my current research interests hover around Differential Privacy, Federated Learning, Decentralised Learning, and, in general, Privacy-Preserving Machine Learning. Occasionally, I also get fascinated by (and end up working on) Location Privacy.

I study, talk, and do mathematics most of the time; when not, I am typically immersed in chess, cricket, puzzles, or stand-up comedy. Of late, I have got myself back into the habit of (non-academic) reading in my leisure and learning about (often, apparently, useless) facts from across the realms of life. I am trying to learn French with limited success so far. The list of some of my current favourite results and theorems in mathematics (a list that, obviously, everyone should maintain) include:

• The limit of the derangement probability of a set as its size approaches infinity is $e^{-1}$
• Gershgorin circle theorem
• Sleeping Beauty problem
• Banach–Tarski paradox
• “A Drunk Man Will Find His Way Home but a Drunk Bird May Get Lost Forever”

Publications

Peer-reviewed conferences

• (2023) Sayan Biswas, Kangsoo Jung, Catuscia Palamidessi: “Tight Differential Privacy Guarantees for the Shuffle Model with k-Randomized Response”. Proceedings of the 16th International Symposium on Foundations and Practice of Security (FPS) 2023, pp 440-458, LNCS 14551, Springer. Published: April 25, 2024.   PDF , DOI: 10.1007/978-3-031-57537-2_27


• (2023) Sayan Biswas, Catuscia Palamidessi: “PRIVIC: A privacy-preserving method for incremental collection of location data”. Proceedings on Privacy Enhancing Technologies (PoPETs), Volume 2024, Issue 1, pp 582-596. Published: October, 2023. PDF , DOI: 10.56553/popets-2024-0033


• (2023) Filippo Galli, Sayan Biswas, Kangsoo Jung, Tommaso Cucinotta, Catuscia Palamidessi: “Group privacy for Personalized Federated Learning”. Proceedings of the 9th International Conference on Information Systems Security and Privacy (ICISSP) 2023, pp 252-263, SciTePress Digital Library. Published: February 24, 2023. PDF , DOI: 10.5220/0011885000003405


• (2022) Sayan Biswas, Graham Cormode, Carsten Maple: “Impact of Sampling on Locally Differentially Private Data Collection”. Proceedings of the 8th Competitive Advantage in the Digital Economy -- Resilience, Sustainability, Responsibility, and Identity (CADE) 2022, pp 64-70, IET Digital Library and IEEE Xplore. Published: November 9, 2022. Winner of the Best Paper Award.   PDF , DOI: 10.1049/icp.2022.2042


• (2022) Sayan Biswas, Kangsoo Jung, Catuscia Palamidessi: “Tight Differential Privacy Blanket for Shuffle Model”. Proceedings of the 8th Competitive Advantage in the Digital Economy -- Resilience, Sustainability, Responsibility, and Identity (CADE) 2022, pp 61-63, IET Digital Library and IEEE Xplore. Published: November 9, 2022.   PDF , DOI: 10.1049/icp.2022.2041


• (2021) Sayan Biswas, Kangsoo Jung, Catuscia Palamidessi: “An Incentive Mechanism for Trading Personal Data in Data Markets”. Proceedings of the 18th International Colloquium on Theoretical Aspects of Computing (ICTAC) 2021, pp 197-213, LNCS 12819, Springer. Published: August 20, 2021.   PDF , DOI: 10.1007/978-3-030-85315-0_12

Journals

• (2024) Ugur Ilker Atmaca, Sayan Biswas, Carsten Maple, Catuscia Palamidessi: “A Privacy-Preserving Querying Mechanism with High Utility for Electric Vehicles”. IEEE Open Journal of Vehicular Technology, Volume 5, pp 262-277. Published: January 30, 2024.   PDF , DOI: 10.1109/OJVT.2024.3360302


• (2023) Filippo Galli, Kangsoo Jung, Sayan Biswas, Catuscia Palamidessi, Tommaso Cucinotta: “Advancing Personalized Federated Learning: Group Privacy, Fairness, and Beyond”. Springer Nature Computer Science, Volume 4, Issue 6, Article 831 (2023). Published: October 28, 2023.   PDF , DOI: 10.1007/s42979-023-02292-0

Book sections

• (2021) Kangsoo Jung, Sayan Biswas, Catuscia Palamidessi: “Establishing the Price of Privacy in Federated Data Trading”. Protocols, Strands, and Logic, pp 232-250, LNCS 13066, Springer. Published: November 19, 2021.   PDF , DOI: 10.1007/978-3-030-91631-2_13

Non-archival workshops

• (2023) Sayan Biswas, Catuscia Palamidessi: “PRIVIC: A privacy-preserving method for incremental collection of location data”. Theory and Practice of Differential Privacy (TPDP) 2023. September 27-28, 2023; Boston, USA.


• (2023) Filippo Galli, Sayan Biswas, Kangsoo Jung, Tommaso Cucinotta, Catuscia Palamidessi: “On the adaptive sensitivity of differentially private machine learning”. The 4th Workshop on Privacy-Preserving Artificial Intelligence in Conjunction with AAAI (PPAI) 2023. February 13, 2023; Washington DC, USA.


• (2022) Filippo Galli, Sayan Biswas, Kangsoo Jung, Tommaso Cucinotta, Catuscia Palamidessi: “Group privacy for Personalized Federated Learning”. International Workshop on Federated Learning: Recent Advances and New Challenges in Conjunction with NeurIPS (FL-NeurIPS) 2022. December 2, 2022; New Orleans, USA. One of the 12 amongst the 103 submissions selected for oral presentation.  Workshop paper

News and Updates

• Oct'23: I am moving to Lausanne, Switzerland and am starting as a Postdoctoral Researcher at the SaCS lab at EPFL under the supervision of Anne-Marie Kermarrec to work on designing privacy-preserving decentralised models.
  


• Oct'23: I successfully defended my PhD thesis titled Understanding and optimizing the trade-off between privacy and utility from a foundational perspective and have been awarded the title of "Doctor of Computer Science" from École Polytechnique and Institut Polytechnique de Paris!
  – Photo ft. the moment of the official change from "Mr. Sayan Biswas" to "Dr. Sayan Biswas".


• Sep'23: Recently, I was visiting Annabelle McIiver and Natasha Fernandes at at Macquarie University in Sydney, Australia, studying the information leakage from differentially private mechanisms from a foundational perspective and doing lots of fun maths!
  


• Aug'23: Continuing its journey of success, our paper "PRIVIC: A privacy-preserving method for incremental collection of location data" got selected for presentation at the Theory and Practice of Differential Privacy Workshop (TPDP) 2023 to be held in Boston, USA on September 27-28, 2023.
  


• Aug'23: Our paper "PRIVIC: A privacy-preserving method for incremental collection of location data" got accepted for publication at the 24th Privacy Enhancing Technologies Symposium (PETS 2024) to be held in Bristol, UK in July, 2024.
  


• Jul'23: Presented an ongoing work "Characterizing the information leakage from gradient updates in federated learning" at the July seminar by FedMalin held online on July 10, 2023.
  – Photo illustrating the philosophy "When someone asks a question during your talk, first have some tea."


• Feb'23: Presented our position paper "On the adaptive sensitivity of differentially private machine learning" at the 4th AAAI Workshop on Privacy-Preserving Artificial Intelligence (PPAI-23) held in conjunction with AAAI 2023 in Washington DC, USA on February 13, 2023.
  


• Dec'22: Presented our paper "Group privacy for personalized federated learning" at the International Workshop on Federated Learning: Recent Advances and New Challenges in Conjunction with NeurIPS (FL-NeurIPS) 2022 in New Orleans, USA on December 2, 2022.
  


• Nov'22: Presented our paper "Group privacy for personalized federated learning" at the Privacy Preserving Machine Learning (PPML) Workshop organised by Meta, Paris on November 9, 2022.
  – Photo ft. three of the five co-authors of the work who struggle to pose elegantly.


• Jun'22: Gave a talk on "A privacy-preserving method for incremental collection of location data: Differential Privacy and beyond" at the 7th Franco-Japanese Cybersecurity Workshop in Tokyo, Japan on October 24, 2022.
  – Photo ft. an underslept, overtravelled, and perpetually excited Sayan giving the talk.


• Aug'22: Designed and delivered a hands-on session to implement differential privacy with federated learning at the 2nd INRIA-DFKI European Summer School on AI (IDESSAI 2022) held in Saarbrücken, Germany from August 29 to September 2, 2022.
  – The course got a special mention as the most liked course taught in the summer school. :)


• Aug'22: Won the "Best Paper Award" for the paper "Impact of Sampling on Locally Differentially Private Data Collection" at CADE 2022 held in Venice, Italy on June 13-15, 2022.
  – Certificate conferring the award.


• Jun'22: Presented an on-going work on "Three-way optimization of privacy and utility of location data" at APVP 2022 in Chatenay sur Seine, France on June 15, 2022.
  – Photo ft. the most Baroque setting and the most non-Baroque outfit to give a talk in.


• Jun'22: Presented the paper "Impact of Sampling on Locally Differentially Private Data Collection" at CADE 2022 in Venice, Italy on June 13-15, 2022.
  – Photo ft. an elated Sayan engrossed in explaining differential privacy to the crowd.