Lezioni di Francisco Paco Villaescusa

Lezioni di Francisco Paco Villaescusa

di ALEXANDRO SARO -
Numero di risposte: 0

Cari studenti, sperando di farvi cosa gradita vi segnalo un seminario all'osservatorio e queste due lezioni che terrà prossimamente il Dott. Francisco Paco Villaescusa. Paco viene come nostro ospite visiting scientist di Dipartimento da NY dove lavora come ricercatore presso il Flat Iron Institute ed è esperto rinomato di ML.


Tuesday, 17 June11:30am – 12:30pm

OATS extra seminar: join in person in the Library at Castello Basevi or remotely on meet.google.com/zfy-mtgs-tpc

Title: Astronomy in the era of AI multi-agent systems

Abstract: I will present a new AI multi-agent system designed to carry out end-to-end scientific research. I will show how, with a relatively general description of some data (e.g. numerical simulations), the system can generate hypotheses, develop research plans, write and execute code, make plots, and write full scientific papers. I will try to generate several papers during the seminar to show its capabilities. I will discuss how these systems can accelerate scientific research massively, but I will also discuss the perils and ethical questions that such a system raises. 



"Tackling complex problems in cosmology with machine learning"
Francisco Paco Villaescusa
Flatiron Institute, New York (U.S.)

Thursday 19/6/2025 11:00
Aula A, Dipartimento di Fisica
Live streaming: Teams link

In this lecture, I will describe a set of complex problems in cosmology that have impeded the community from making progress. I will then describe how machine learning methods have been developed to tackle these problems.

"AI Agents: a new paradigm for science?
Francisco Paco Villaescusa
Flatiron Institute, New York (U.S.)

Friday 20/6/2025 14:00
Aula A, Dipartimento di Fisica
Live streaming: Teams link

In this lecture, I will introduce AI agents. I will show what they are, their capabilities, and their role in science. I will illustrate how to build a simple AI agent in python and then explain how complex multi-agent systems are built and deployed. I will illustrate the capabilities of these systems by showing how to write a scientific paper live (bring your own laptop if possible).