Samuel Maddock
About Me
I am a final-year PhD student supervised by Prof. Graham Cormode and Prof. Carsten Maple (WMG). Before starting my PhD I graduated with a first-class MEng in Discrete Mathematics here at the University of Warwick.
During the Summer of 2022, I was a research intern at Meta AI in Paris. I was supervised by Pierre Stock and Alexandre Sablayrolles working on empirical privacy measurement in the federated setting.
During the Summer of 2025, I was a research scientist intern at Meta AI in Menlo Park, California. I was part of the Central Applied Science (CAS) team and supervised by Shripad Gade.
From June 2024-May 2025 & Sept 2025-Dec 2025 I have been a part-time research collaborator at Meta AI working on problems within private synthetic data and tabular generative modelling.
Research Interests
My research interests lie at the intersection of computer security, data privacy and machine learning. I am mainly interested in the area of privacy-preserving machine learning (PPML) which seeks to develop ML algorithms to learn privately and securely from data which is often highly distributed. This includes areas such as Differential Privacy (DP), Federated Learning (FL) and Federated Analytics.
Conference Publications
- Samuel Maddock, Graham Cormode, Carsten Maple. "FLAIM: AIM-based Synthetic Data Generation in the Federated Setting", In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD '24) (Slides, Poster)
- Samuel Maddock, Alexandre Sablayrolles and Pierre Stock. "CANIFE: Crafting Canaries for Empirical Privacy Measurement in Federated Learning". The Eleventh International Conference on Learning Representations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023. (Slides, Poster)
- Samuel Maddock, Graham Cormode, Tianhao Wang, Carsten Maple and Somesh Jha. "Federated Boosted Decision Trees with Differential Privacy". In Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security (CCS ’22), November 7–11, 2022, Los Angeles, CA, USA. (Slides)
- Graham Cormode, Samuel Maddock, and Carsten Maple. "Frequency estimation under local differential privacy." Proceedings of the VLDB Endowment 14.11 (2021): 2046-2058. (Slides , Poster)
Workshop Papers
- [Spotlight] Samuel Maddock, Shripad Gade, Graham Cormode, Will Bullock, "Leveraging Vertical Public-Private Split for Improved Synthetic Data Generation", SynthData Workshop @ ICLR 2025
Preprints
- Samuel Maddock, Graham Cormode, Carsten Maple. "Private Federated Multiclass Post-hoc Calibration", arXiv
Accepted Tutorials
- Tutorial @ KDD 2025 Toronto, Canada: Synthetic Tabular Data: methods, attacks and defenses, Graham Cormode, Samuel Maddock, Enayat Ullah, Shripad Gade,
- Tutorial @ VLDB 2025 London, UK: Synthetic Tabular Data: methods, attacks and defenses, Graham Cormode, Samuel Maddock, Enayat Ullah, Shripad Gade,
Invited Talks & Service
- Trustworthy Synthetic Data in Practice (TSDiP) @ Alan Turing Institute (2024)
- WPCCS'23 (Awarded 'Best in Session')
- FLOW Seminar: CANIFE (2023)
- Google FL Workshop (2022)
- Invited Reviewer: KDD 2024, NeurIPS 2024, ICLR 2025, ICML 2025, NeurIPS 2025
Teaching
I have been a teaching assistant for the following modules:
