Machine Learning for Simulation meeting

Thursday 30 Sept 2021, 17:00 → 18:30 Europe/Zurich

Lecture on Transport methods for simulation-based inference by Prof.Youssef Marzouk, Massachusetts Institute of Technology (MIT).

 

Abstract

 

Many practical Bayesian inference problems fall into the "likelihood-free" setting, where evaluations of the likelihood function or prior density are unavailable or intractable; instead one can only simulate (i.e., draw samples from) the associated distributions. I will discuss how transportation of measure can help solve such problems, by constructing maps that push prior samples, or samples from a joint parameter-data prior, to the desired conditional distribution. These methods have broad utility for inference in stochastic and generative models, as well as for sequential data assimilation problems. Key issues in this construction center on: (1) the estimation of transport maps from few samples; and (2) parameterizations of monotone maps. I will discuss developments on both fronts, including some recent efforts in joint dimension reduction for conditional sampling in high dimensions.

 

This is joint work with Ricardo Baptista, Olivier Zahm, and Jakob Zech.

 

Join Zoom Meeting
https://cern.zoom.us/j/66228732578?pwd=ZFpUb1krZ1hGMGhISTMzNnZFWFFDZz09

Meeting ID: 662 2873 2578
Passcode: 10082367
One tap mobile
+33170379729,,66228732578# France
+33186995831,,66228732578# France

Dial by your location
        +33 1 7037 9729 France
        +33 1 8699 5831 France
        +33 1 7037 2246 France
        +41 43 210 70 42 Switzerland
        +41 43 210 71 08 Switzerland
        +41 31 528 09 88 Switzerland
Meeting ID: 662 2873 2578
Find your local number: https://cern.zoom.us/u/cxLZ1B4g

Join by SIP
66228732578@188.184.89.188
66228732578@188.185.118.153

Join by H.323
188.184.89.188
188.185.118.153
Meeting ID: 662 2873 2578
Passcode: 10082367

17:00 → 18:00 Transport methods for simulation-based inference

Speaker: Prof. Youssef Marzouk (MIT)

cern-ymarz.pdf

Recording

18:00 → 18:30 Q&A