Inaugural lecture: "Centaur Science: Particle Physics meets Machine Learning"

• Jesse Thaler (MIT)

Room: CYCL 01 Modern machine learning has had an outsized impact on many scientific fields, and particle physics is no exception. What is special about particle physics, though, is the vast amount of theoretical knowledge that we already have about many problems in the field, as well as the daunting deluge of data coming from flagship experiments like the Large Hadron Collider (LHC). In this lecture, I will explain how one can teach a machine to "think like a physicist" by embedding theoretical principles into advanced machine learning architectures. At the same time, I will advocate that physicists must learn how to "think like a machine" to maximize the physics reach of the LHC. These joint developments are leading to a new kind of "centaur science" that, analogously to the mythical centaur, draws half from particle physics and half from machine learning.

Lecture 2: "Machine Learning through the Lens of Physics"

• Jesse Thaler (MIT)

Room: SUD 01 Machine learning has a reputation for yielding "black box" solutions that cannot be properly interpreted. In this lecture, I explain foundational aspects of machine learning in a language familiar to physicists and argue that interpretability must be assessed with concrete physics goals in mind.

Public lecture: "Deep learning + deep thinking = deeper understanding"

• Jesse Thaler (MIT)

Room: SUD 09 <img src="https://cern.ch/durieux/lemaitre-chair-banner-p2-v2.png" style="width:900px;"/> Artificial intelligence is transforming many aspects of society, including the ways that scientists are pursuing groundbreaking discoveries. This public lecture will highlight ways that "physics intelligence" can be incorporated into artificial intelligence, with the twin goals of answering fundamental questions about the universe and illuminating the mechanisms behind machine learning.

Lecture 3: "Frequent(ist)ly Asked Questions"

• Jesse Thaler (MIT)

Room: SUD 01 Most particle physics results are framed in the language of frequentist inference, whereas machine learning fits more naturally into a Bayesian framework. In this lecture, I explain how to extract frequentist information from neural networks and why we must incorporate rigorous statistical uncertainties into machine learning pipelines.

Lecture 4: "Jets in Space"

• Jesse Thaler (MIT)

Room: SUD 11 Optimal transport (OT) is a fascinating branch of mathematics which can be used to define a useful notion of geometry for datasets. In this lecture, I explain how many concepts and techniques from particle physics have a natural OT interpretation, particularly when describing jets of particles produced from the strong force.

Lecture 5: "The Future of AI+Physics"

• Jesse Thaler (MIT)

Room: SUD 01 The practice of physics is shaped not only by the questions we ask but by the tools at our disposal. In the era of artificial intelligence (AI), what will physics research look like? I don't know, but in this lecture, I will highlight some of the ongoing trends in AI+Physics and speculate how the field might evolve.