Srinivas Devadas
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Srinivas Devadas has a total of 107 co-authors. Here are the top co-authors by number of shared publications:
- Hanshen Xiao
Purdue University, West Lafayette, IN, USA and NVIDIA, West Lafayette, IN, USA
7 shared publications - Marten van Dijk
Massachusetts Institute of Technology, Cambridge, MA 02139
5 shared publications - Omer Khan
University of Connecticut, Storrs, CT, USA
5 shared publications - Blaise Gassend
Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
3 shared publications - Dwaine Clarke
University of the West Indies
3 shared publications - Jun Wan
Five Rings, New York, NY, USA
3 shared publications
Latest Publications
- Open Access
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Enabling AI ASICs for Zero Knowledge Proof
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessEnabling AI ASICs for Zero Knowledge Proof
- Open Access
PAC-Private Responses with Adversarial Composition
Modern machine learning models are increasingly deployed behind APIs. This renders standard weight-privatization methods (e.g. DP-SGD) unnecessarily noisy at the cost of utility. While model weights may vary significantly across training datasets, model responses to specific inputs are much lower dimensional and more stable. This motivates enforcing privacy guarantees directly on model outputs. We approach this under PAC privacy, which provides instance-based privacy guarantees for arbitrary black-box functions by controlling mutual information (MI). Importantly, PAC privacy explicitly rewards output stability with reduced noise levels. However, a central challenge remains: response privacy requires composing a large number of adaptively chosen, potentially adversarial queries issued by untrusted users, where existing composition results on PAC privacy are inadequate. We introduce a new algorithm that achieves adversarial composition via adaptive noise calibration and prove that mutual information guarantees accumulate linearly under adaptive and adversarial querying. Experiments across tabular, vision, and NLP tasks show that our method achieves high utility at extremely small per-query privacy budgets. On CIFAR-10, we achieve 87.79% accuracy with a per-step MI budget of $2^{-32}$. This enables serving one million queries while provably bounding membership inference attack (MIA) success rates to 51.08% -- the same guarantee of $(0.04, 10^{-5})$-DP. Furthermore, we show that private responses can be used to label public data to distill a publishable privacy-preserving model; using an ImageNet subset as a public dataset, our model distilled from 210,000 responses achieves 91.86% accuracy on CIFAR-10 with MIA success upper-bounded by 50.49%, which is comparable to $(0.02,10^{-5})$-DP.Open AccessPAC-Private Responses with Adversarial Composition
Modern machine learning models are increasingly deployed behind APIs. This renders standard weight-privatization methods (e.g. DP-SGD) unnecessarily noisy at the cost of utility. While model weights may vary significantly across training datasets, model responses to specific inputs are much lower dimensional and more stable. This motivates enforcing privacy guarantees directly on model outputs. We approach this under PAC privacy, which provides instance-based privacy guarantees for arbitrary black-box functions by controlling mutual information (MI). Importantly, PAC privacy explicitly rewards output stability with reduced noise levels. However, a central challenge remains: response privacy requires composing a large number of adaptively chosen, potentially adversarial queries issued by untrusted users, where existing composition results on PAC privacy are inadequate. We introduce a new algorithm that achieves adversarial composition via adaptive noise calibration and prove that mutual information guarantees accumulate linearly under adaptive and adversarial querying. Experiments across tabular, vision, and NLP tasks show that our method achieves high utility at extremely small per-query privacy budgets. On CIFAR-10, we achieve 87.79% accuracy with a per-step MI budget of $2^{-32}$. This enables serving one million queries while provably bounding membership inference attack (MIA) success rates to 51.08% -- the same guarantee of $(0.04, 10^{-5})$-DP. Furthermore, we show that private responses can be used to label public data to distill a publishable privacy-preserving model; using an ImageNet subset as a public dataset, our model distilled from 210,000 responses achieves 91.86% accuracy on CIFAR-10 with MIA success upper-bounded by 50.49%, which is comparable to $(0.02,10^{-5})$-DP. - 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Proceedings of the IEEE
A Tutorial on Oblivious Random Access Memory
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Proceedings of the IEEE
A Tutorial on Oblivious Random Access Memory
- Open Access
- 1 January 2026
- Published by Privacy Enhancing Technologies Symposium Advisory Board in Proceedings on Privacy Enhancing Technologies
- Vol. 2026 (1) , 450-469
Making Sense of Private Advertising: A Principled Approach to a Complex Ecosystem
In this work, we model the end-to-end pipeline of the advertising ecosystem, allowing us to identify two main issues with the current trajectory of private advertising proposals. First, prior work has largely considered ad targeting and engagement metrics individually rather than in composition. This has resulted in privacy notions that, while reasonable for each protocol in isolation, fail to compose to a natural notion of privacy for the ecosystem as a whole, permitting advertisers to extract new information about the audience of their advertisements. The second issue serves to explain the first: we prove that perfect privacy is impossible for any, even minimally, useful advertising ecosystem, due to the advertisers' expectation of conducting market research on the results. Having demonstrated that leakage is inherent in advertising, we re-examine what privacy could realistically mean in advertising, building on the well-established notion of sensitive data in a specific context. We identify that fundamentally new approaches are needed when designing privacy-preserving advertising subsystems in order to ensure that the privacy properties of the end-to-end advertising system are well aligned with people's privacy desires.
- 1 January 2026
- Published by Privacy Enhancing Technologies Symposium Advisory Board in Proceedings on Privacy Enhancing Technologies
- Vol. 2026 (1) , 450-469
Open AccessMaking Sense of Private Advertising: A Principled Approach to a Complex Ecosystem
In this work, we model the end-to-end pipeline of the advertising ecosystem, allowing us to identify two main issues with the current trajectory of private advertising proposals. First, prior work has largely considered ad targeting and engagement metrics individually rather than in composition. This has resulted in privacy notions that, while reasonable for each protocol in isolation, fail to compose to a natural notion of privacy for the ecosystem as a whole, permitting advertisers to extract new information about the audience of their advertisements. The second issue serves to explain the first: we prove that perfect privacy is impossible for any, even minimally, useful advertising ecosystem, due to the advertisers' expectation of conducting market research on the results. Having demonstrated that leakage is inherent in advertising, we re-examine what privacy could realistically mean in advertising, building on the well-established notion of sensitive data in a specific context. We identify that fundamentally new approaches are needed when designing privacy-preserving advertising subsystems in order to ensure that the privacy properties of the end-to-end advertising system are well aligned with people's privacy desires.
- Open Access
Making Sense of Private Advertising: A Principled Approach to a Complex Ecosystem
In this work, we model the end-to-end pipeline of the advertising ecosystem, allowing us to identify two main issues with the current trajectory of private advertising proposals. First, prior work has largely considered ad targeting and engagement metrics individually rather than in composition. This has resulted in privacy notions that, while reasonable for each protocol in isolation, fail to compose to a natural notion of privacy for the ecosystem as a whole, permitting advertisers to extract new information about the audience of their advertisements. The second issue serves to explain the first: we prove that \textit{perfect} privacy is impossible for any, even minimally, useful advertising ecosystem, due to the advertisers' expectation of conducting market research on the results. Having demonstrated that leakage is inherent in advertising, we re-examine what privacy could realistically mean in advertising, building on the well-established notion of \textit{sensitive} data in a specific context. We identify that fundamentally new approaches are needed when designing privacy-preserving advertising subsystems in order to ensure that the privacy properties of the end-to-end advertising system are well aligned with people's privacy desires.
Open AccessMaking Sense of Private Advertising: A Principled Approach to a Complex Ecosystem
In this work, we model the end-to-end pipeline of the advertising ecosystem, allowing us to identify two main issues with the current trajectory of private advertising proposals. First, prior work has largely considered ad targeting and engagement metrics individually rather than in composition. This has resulted in privacy notions that, while reasonable for each protocol in isolation, fail to compose to a natural notion of privacy for the ecosystem as a whole, permitting advertisers to extract new information about the audience of their advertisements. The second issue serves to explain the first: we prove that \textit{perfect} privacy is impossible for any, even minimally, useful advertising ecosystem, due to the advertisers' expectation of conducting market research on the results. Having demonstrated that leakage is inherent in advertising, we re-examine what privacy could realistically mean in advertising, building on the well-established notion of \textit{sensitive} data in a specific context. We identify that fundamentally new approaches are needed when designing privacy-preserving advertising subsystems in order to ensure that the privacy properties of the end-to-end advertising system are well aligned with people's privacy desires.
- Open Access
- 19 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 4214-4228
One-Sided Bounded Noise: Theory, Optimization Algorithms and Applications
- 19 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 4214-4228
Open AccessOne-Sided Bounded Noise: Theory, Optimization Algorithms and Applications
- Open Access
- 17 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 300-307
Remote Direct Code Execution
- 17 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 300-307
Open AccessRemote Direct Code Execution
- 12 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Security & Privacy
- Vol. 23 (4) , 2-7
PAC Privacy and Black-Box Privatization
This article discusses a new privacy definition called probably approximately correct (PAC) Privacy, which offers a formal probabilistic characterization of privacy risk aligned with general privacy concerns. PAC Privacy enables automated, universal, and provable privatization for any (possibly black-box) computation.
- 12 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Security & Privacy
- Vol. 23 (4) , 2-7
PAC Privacy and Black-Box Privatization
This article discusses a new privacy definition called probably approximately correct (PAC) Privacy, which offers a formal probabilistic characterization of privacy risk aligned with general privacy concerns. PAC Privacy enables automated, universal, and provable privatization for any (possibly black-box) computation.
- Open Access
- 15 July 2025
- Published by Association for Computing Machinery (ACM) in Journal of the ACM
Expected Constant Round Byzantine Broadcast under Dishonest Majority
Byzantine Broadcast (BB) is a central question in distributed systems, and an important challenge is to understand its round complexity. Under the honest majority setting, it is long known that there exist randomized protocols that can achieve BB in expected constant rounds, regardless of the number of nodes n . However, whether we can match the expected constant round complexity in the corrupt majority setting — or more precisely, when f ≥ n /2 + ω (1) — remains unknown, where f denotes the number of corrupt nodes. In this paper, we are the first to resolve this long-standing question. We show how to achieve BB in expected O (( n /( n − f )) 2 ) rounds. Our results hold under a weakly adaptive adversary who cannot perform “after-the-fact removal” of messages already sent by a node before it becomes corrupt. We also assume trusted setup and the Decision Linear (DLIN) assumption in bilinear groups.
- 15 July 2025
- Published by Association for Computing Machinery (ACM) in Journal of the ACM
Open AccessExpected Constant Round Byzantine Broadcast under Dishonest Majority
Byzantine Broadcast (BB) is a central question in distributed systems, and an important challenge is to understand its round complexity. Under the honest majority setting, it is long known that there exist randomized protocols that can achieve BB in expected constant rounds, regardless of the number of nodes n . However, whether we can match the expected constant round complexity in the corrupt majority setting — or more precisely, when f ≥ n /2 + ω (1) — remains unknown, where f denotes the number of corrupt nodes. In this paper, we are the first to resolve this long-standing question. We show how to achieve BB in expected O (( n /( n − f )) 2 ) rounds. Our results hold under a weakly adaptive adversary who cannot perform “after-the-fact removal” of messages already sent by a node before it becomes corrupt. We also assume trusted setup and the Decision Linear (DLIN) assumption in bilinear groups.
- 8 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Computer
- Vol. 58 (1) , 105-108
Toward a Universal Cryptographic Accelerator
Cyberphysical systems have disseminated devices that can be untrustworthy or compromised. Nevertheless, the privacy and integrity of computation and data can be guaranteed through cryptographic protocols. We address the computational burden posed by cryptography, and argue for a synergistic approach of designing programmable hardware accelerators for cryptography, followed by tailoring cryptographic protocols to this hardware.
- 8 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Computer
- Vol. 58 (1) , 105-108
Toward a Universal Cryptographic Accelerator
Cyberphysical systems have disseminated devices that can be untrustworthy or compromised. Nevertheless, the privacy and integrity of computation and data can be guaranteed through cryptographic protocols. We address the computational burden posed by cryptography, and argue for a synergistic approach of designing programmable hardware accelerators for cryptography, followed by tailoring cryptographic protocols to this hardware.