The skewed Bernstein-von Mises theorem is online arxiv.org/abs/2301.03038. With Francesco Pozza and Botond Szabo we derive a new limiting law given by tractable generalized skew-normals that remarkably improves the convergence rate (and practical accuracy) of the classical BvM result!

🤩 Congratulations to my former PhD student Francesco Pozza who received the Laplace award 🏆 at #JSM2024, for the article "skewed Bernstein-von Mises theorem and skew-modal approximations" (arxiv.org/abs/2301.03038) (recently accepted for publication in the Annals of Statistics ‼️)

Our paper "Concentration of discrepancy-based approximate Bayesian computation via Rademacher complexity" has been accepted by the Annals of Statistics! 🎉🎉🎉 Thanks to my coauthors Daniele Durante and @PierreAlquier for sharing this amazing journey! 🙏 arxiv.org/abs/2206.06991

😎 For any, already-derived, symmetric approximation of a generic posterior distribution, we provide, at no additional optimization costs, a similarly-tractable, yet provably more accurate, skew-symmetric approximation! (arxiv.org/abs/2409.14167) (with Francesco Pozza and Botond Szabo)