Bayesian model averaging with exact inference of likelihood- free scoring rule posteriors [23/01/2024, PariSanté campus]

A special “All about that Bayes” seminar in Paris (PariSanté campus, 23/01, 16:00-17:00) next week by my Warwick collegue and friend Rito:

Bayesian Model Averaging with exact inference of likelihood- free Scoring Rule Posteriors

Ritabrata Dutta, University of Warwick

A novel application of Bayesian Model Averaging to generative models parameterized with neural networks (GNN) characterized by intractable likelihoods is presented. We leverage a likelihood-free generalized Bayesian inference approach with Scoring Rules. To tackle the challenge of model selection in neural networks, we adopt a continuous shrinkage prior, specifically the horseshoe prior. We introduce an innovative blocked sampling scheme, offering compatibility with both the Boomerang Sampler (a type of piecewise deterministic Markov process sampler) for exact but slower inference and with Stochastic Gradient Langevin Dynamics (SGLD) for faster yet biased posterior inference. This approach serves as a versatile tool bridging the gap between intractable likelihoods and robust Bayesian model selection within the generative modelling framework.

A. K. Md. Ehsanes Saleh (01 Jan 1932 – 03 Sept 2023)

Just learned this day that Professor A. K. Md. Ehsanes Saleh passed away in early September. I first met him sometimes in the Fall of 1987, while visiting (from Purdue where I was visiting professor) my wife in Ottawa (where she was pursuing a Master in Electrical Engineering). I knew of his papers on shrinkage and pre-test estimators and dropped by Carleton University, where he taught and worked most of his life, for a casual talk. He was incredibly welcoming and friendly to an unknown junior researcher who had dropped by with no warning on a Friday afternoon. We then kept in touch about research projects and he made me an offer to visit Carleton over the Summer of 1988, with a welcome financial support that allowed us to rent a better lodging by the University of Ottawa (which my wife kept for the following year). This suited me most perfectly as I could spend the summer (May-August) with my wife and work with Professor Saleh on shrinkage topics, which was most enjoyable (if not immensely innovative), although the move involved a non-stop 14h drive from West Lafayette to Ottawa! The whole group of statisticians and probabilists at Carleton was unbelievably friendly as well and contributed, along with the stressless atmosphere of the Canadian capital and the endless nearby parks, to make that summer of 1988 a fabulous one. We renewed the experiment the following summer of 1989, when I left Cornell at the end of their semester, again a great one, when I also met Tatsuya Kubokawa who was visiting Professor Saleh as well. After those two years, I had very few opportunities to visit Ottawa and hence to meet him again, even though I remember having lunch with him at a Franco-Canadian meeting in 2008. I do and will remember him as a humble and selfless man, despite his accomplishments of being the first Bangladeshi statistician in receiving many awards and distinctions, always amicable and full of tolerance and helpful advice.

35 years ago…

Bill’s 80th birthday

estimation of a normal mean matrix

A few days ago, I noticed the paper Estimation under matrix quadratic loss and matrix superharmonicity by Takeru Matsuda and my friend Bill Strawderman had appeared in Biometrika. (Disclaimer: I was not involved in handling the submission!) This is a “classical” shrinkage estimation problem in that covariance matrix estimators are compared under under a quadratic loss, using Charles Stein’s technique of unbiased estimation of the risk is derived. The authors show that the Efron–Morris estimator is minimax. They also introduce superharmonicity for matrix-variate functions towards showing that generalized Bayes estimator with respect to a matrix superharmonic priors are minimax., including a generalization of Stein’s prior. Superharmonicity that relates to (much) earlier results by Ed George (1986), Mary-Ellen Bock (1988),  Dominique Fourdrinier, Bill Strawderman, and Marty Wells (1998). (All of whom I worked with in the 1980’s and 1990’s! in Rouen, Purdue, and Cornell). This paper also made me realise Dominique, Bill, and Marty had published a Springer book on Shrinkage estimators a few years ago and that I had missed it..!