Events

A unified, decision-theoretic view of machine learning

Speaker: Kevin P. Murphy

Location: 60 Fifth Avenue, Room 7th floor common area

Date: Tuesday, May 2, 2023

I will present a unified decision-theoretic perspective on the field of machine learning, following the structure of my recent book, "Probabilistic Machine Learning: Advanced Topics". This is centered on what I call the "4 pillars of ML": one-shot decisions (prediction), sequential decisions (control), decisions with a latent loss (discovery), and decisions with ambiguous loss (generation).  For each of these tasks, I will give a brief overview of some common techniques, including a few of my own contributions, and mention open problems.