Experimental Particle and Astro-Particle Physics Seminar

Nicola Serra (Universitaet Zuerich (CH))

The goal of this talk is to focus on generative models that are not so popular in the particle physics community. While GANs and AEs represent the state of the art in unsupervised learning, their training is typically difficult and resource intensive. In this talk we will introduce Restricted Boltzmann Machines (RBMs) and showcase their main advantages (such as ease of training and efficient conditional sampling) which may make them better suited than their competitors for some applications. The models will be introduced from scratch and two recent empirical applications (to credit risk management and pharmaceutical product liability) coming from my research will be presented.

Zoom room link

Meeting ID: 912 1383 7955
Passcode: 069577


    • 2:00 PM 3:00 PM
      Restricted Boltzmann Machines: theory and applications 1h
      Speaker: Gabriele Visentin (University of Zurich)