Tomorrow - that is, November 8th, at 8AM GMT - I am chairing a session titled "Artificial Intelligence for Physics Research, and Physics Research for Artificial Intelligence" at the Vth USERN Congress. The event takes place in Tehran, and is broadcast via zoom. If you are interested in the talks, of which I give some detail below, you will be able to connect through this link.
The agenda of the workshop is as follows (those shown are are Tehran times):

11.30-11.55 Tommaso Dorigo, “Artificial intelligence and fundamental physics research”

12.00-12.25 André David, “How we discovered the Higgs ahead of schedule: ML’s role in unveiling the keystone of elementary particle physics”

12.30-12.55 Andrey Ustyuzhanin, “Artificial Intelligence and Optimization Challenges in Physical Sciences”

13.00-13.25 Pietro Vischia, “Pattern recognition and image reconstruction: from industry to particle physics (and back?)”

13.25-13.30 Tommaso Dorigo - Closing remarks


As you can see, the speakers are all of high level, and they will provide a view of the topic from different angles. I hope some of you will be able to follow the presentations! 


Short abstracts of the presentations are given below:

T. Dorigo: “Artificial intelligence and fundamental physics research”

Abstract: The possibility of creating machines that show traits of intelligent behavior or problem-solving skills was put forth 65 years ago, yet after initial enthusiasm research was dampened by the difficulty of the task and lack of progress. Since the turn of the century the development of machine learning algorithms however provided a number of results and products which could be argued to show intelligence – from language translation, speech and image recognition, to chess-playing neural networks and self-driving vehicles. These successes, which we are growing accustomed to, are leading us to think of Artificial Intelligence (AI) as “whatever machines cannot yet do”. In this presentation the status of AI research and its future will be discussed in the context of its effects on the development of better research in physics.

A. David: “How we discovered the Higgs ahead of schedule: ML’s role in unveiling the keystone of elementary particle physics”

Abstract: In 2010, when the LHC started colliding proton pairs in earnest, multi-variate analyses were newfangled methods starting to make inroads in experimental particle physics. These methods faced widespread skepticism as to their performance and biases, reflecting a winter of suspicion over overtrained neural networks that set in in the late 1990s. Thanks to more robust techniques, like boosted decision trees, it became possible to make better and more extensive use of the full information recorded in particle collisions at the Tevatron and LHC colliders. The Higgs boson discovery by the CMS and ATLAS collaborations in 2012 was only possible because of the use of multi-variate techniques that enhanced the sensitivity by up to the equivalent of having 50% more collision data available for analysis. We will review the use of classification and regression in the Higgs to diphoton search and subsequent discovery, a concrete example of a decade-old ML-based analysis in high-energy particle physics. Particular emphasis will be placed in the modular design of the analysis and the inherent explainability advantages, used to great effect in assuaging concerns raised by hundreds of initially-skeptical colleagues in the CMS collaboration. Finally, we will quickly highlight some particle physics challenges that have contributed to, and made use of, the last decade of graph, adversarial, and deep ML developments.

A. Ustyuzhanin: “Artificial Intelligence and Optimization Challenges in Physical Sciences”

Abstract: In this talk, we provide examples of non-trivial optimization problems arising in various natural science disciplines. Particular emphasis is placed on the problems associated with the use of multiple simulation tools and approaches. The solutions to such challenges rely on the use of generative models to approximate the available observations. Approaches such as Local Generative Surrogate Optimization (L-GSO), Adaptive divergence and probabilistic inference will be described using various properties of the simulator and surrogate models. The solutions to problems of physical detectors of some modern experiments, developed by the Laboratory of Methods for Analysis of Big Data of the Higher School of Economics, are demonstrated.

P. Vischia: "Pattern recognition and image reconstruction: from industry to particle physics (and back?)"

Abstract: In the last decades, machine learning algorithms have been employed to solve a variety of common problems: image recognition algorithms have been useful to automatically sort through postal mail or to automatically recognize faces in social and surveillance networks, and are now commonly used in self-driving cars; path reconstruction is at the core of software that helps you to reach a destination by foot, public transport, or car; text recognition is used to classify and generate new texts starting from known patterns. The same algorithms power the most recent applications of machine learning in particle physics: I will review the major advancements in machine learning algorithms that can serve both society and fundamental research applications.