Adam Smith is a professor of computer science at Boston University, where he leads research at the intersection of differential privacy and cryptography. His work has helped establish the theoretical foundations for privacy-preserving data analysis, influencing both academic research and industrial deployment of private computation systems. Smith is widely cited for his contributions to differential privacy, secure multiparty computation, and the design of algorithms that guarantee formal privacy protections.
Born in the United States, Smith completed his undergraduate studies in mathematics and computer science before earning a PhD in computer science from MIT in 2000. His doctoral research focused on cryptographic protocols and the foundations of secure computation. He subsequently held positions at Carnegie Mellon University and Pennsylvania State University before joining Boston University, where he has been a faculty member since 2013.
Differential Privacy and the Foundations of Private Data Analysis
Smith's early work in the 2000s helped formalize the notion of differential privacy, a framework that quantifies the privacy loss incurred when releasing statistical information about a dataset. He contributed to the development of mechanisms that add calibrated noise to query outputs, ensuring that the presence or absence of any individual record has limited effect on the result. His 2006 paper on the interaction between differential privacy and robust statistics is considered a milestone, demonstrating that private estimators can achieve near-optimal accuracy for a wide range of statistical tasks.
One of Smith's key contributions is the concept of "privacy amplification" through subsampling, which shows that randomly selecting a subset of data before applying a private mechanism can significantly reduce the privacy cost. This principle has become a standard tool in the design of private machine learning algorithms, including those used in deep learning and generative AI systems.
Cryptography and Secure Computation
Beyond differential privacy, Smith has made substantial contributions to cryptography, particularly in the area of secure multiparty computation (MPC). His research has explored how multiple parties can jointly compute a function over their private inputs without revealing those inputs to each other. He has worked on protocols that achieve security against malicious adversaries, improving the efficiency and practicality of MPC for real-world applications.
Smith's cryptographic work also intersects with differential privacy, as he has investigated how cryptographic techniques can be used to enforce privacy guarantees in distributed settings. His collaborative research with Aleksander Madry and others has addressed challenges in private learning and the robustness of machine learning models.
Academic Leadership and Teaching
At Boston University, Smith is affiliated with the BU Center for Information and Systems Engineering and the Hariri Institute for Computing. He teaches graduate courses on privacy, cryptography, and algorithms, mentoring numerous PhD students who have gone on to academic and industry positions. His teaching emphasizes the importance of rigorous formal guarantees in privacy and security, and he has been recognized for his ability to convey complex theoretical concepts to a broad audience.
Smith has served on program committees for major conferences in cryptography and privacy, including CRYPTO, EUROCRYPT, and FOCS. He has also been an associate editor for the Journal of Cryptology and the IEEE Transactions on Information Theory. His editorial work has helped shape the direction of research in privacy-preserving computation.
Impact on Industry and Policy
Smith's research has had a tangible impact beyond academia. His insights into differential privacy have informed the design of privacy-preserving systems at major technology companies, including Apple, Google, and Microsoft. The U.S. Census Bureau adopted differential privacy for the 2020 census, a decision influenced by the theoretical framework that Smith and his colleagues helped develop.
In the policy arena, Smith has advised government agencies on privacy-preserving data sharing and has contributed to reports on the ethical use of data. He has spoken at forums on the balance between data utility and individual privacy, advocating for the adoption of formal privacy guarantees in public data releases.
Selected Awards and Honors
Smith has received several awards for his research, including the ACM SIGACT Distinguished Paper Award and the IEEE Symposium on Security and Privacy Best Paper Award. He is a fellow of the International Association for Cryptologic Research (IACR), recognizing his sustained contributions to the field. His work has been supported by grants from the National Science Foundation and the Defense Advanced Research Projects Agency (DARPA).
As of 2024, Smith continues to lead an active research group at Boston University, exploring new frontiers in privacy-preserving machine learning and secure computation. His ongoing projects include developing private algorithms for large language models and investigating the interplay between differential privacy and adversarial robustness.
References
- Smith, A. (2006). Differential privacy and robust statistics. In Proceedings of the 38th Annual ACM Symposium on Theory of Computing.
- Smith, A., & Thakurta, A. (2013). Differentially private feature selection via stability arguments. In Proceedings of the 30th International Conference on Machine Learning.
- Bun, M., & Smith, A. (2013). Private and efficient data analysis. In Proceedings of the 4th Conference on Innovations in Theoretical Computer Science.