My research is in high-dimensional statistics and machine learning — change-point detection, ranking and crowdsourcing problems, and bandits — and I also work on high-performance / parallel computing on GPUs. Click any card below to dive in; some open interactive animations of the work.
High-Performance Portable GPU Primitives for Arbitrary Types and Operators in Julia
E. Pilliat · 2026 · Preprint
Recovering Labels from Crowdsourced Data: An Optimal and Polynomial-Time Method
E. Pilliat · 2025 · COLT
On Sparsity and Sub-Gaussianity in the Johnson–Lindenstrauss Lemma
A. Garivier · E. Pilliat · 2024 · TMLR
Optimal rates for ranking a permuted isotonic matrix in polynomial time
E. Pilliat · A. Carpentier · N. Verzelen · 2024 · SODA
Optimal permutation estimation in crowd-sourcing problems
E. Pilliat · A. Carpentier · N. Verzelen · 2023 · Annals of Statistics
Optimal multiple change-point detection for high-dimensional data
E. Pilliat · A. Carpentier · N. Verzelen · 2023 · EJS