The "Learning to Discover" workshops and "AI and Physics" conference are taking place at Institut Pascal, a centre set on the top of a hill surrounded by woods near Orsay, France. The event focuses on new artificial intelligence techniques to improve the discovery potential of fundamental science experiments.
Below you can see a summary of the event agenda

- Apr 19-20 Representation learning workshop
- Apr 21-22 Dealing with uncertainties workshop
- Apr 25-26 Generative models workshop

- Apr 27-29 AI and Physics Conference

I attended in person to one of the three workshops last week, which was themed "Dealing with uncertainties". Indeed, the topic of reducing systematic effects on the result of our measurements is one dear to me, as I have published a few years ago (with the main contribution from a former Ph.D. student from the AMVA4NewPhysics network, Pablo de Castro Manzano) a seminal work to attack the problem of including systematic uncertainties in supervised learning problems, to obtain summary statistics that are robust to those biasing effects.

I was surprised by Institut Pascal, which is set on a hill close to Saclay. The countryside is very pleasant, with woods and nice, quiet landscapes; and the building where we are hosted has office space for workshop participants, free coffee, juices and snacks all day long, and is an ideal venue where participants may find ways to entertain scientific discussions and start new collaborations. 

The only catch is that for participants who attend without a car, getting to the venue involves climbing up the hill from the residence where we are lodged. This is not too hard in sunny spring mornings, but I wonder what will happen if we get some bad weather...  Probably I will solve the matter with Uber. I guess I will soon know, as I will fly in again in two days to Orsay for the final conference of the event, where I will be giving a presentation titled "Experiment optimization with differentiable programming" on Thursday morning. Below is the short abstract of my talk:

"In 2012 the imagenet challenge and the discovery of the Higgs boson produced a paradigm shift in particle physics analysis. Today a new revolution awaits to be made. The possibility to map continuously the space of design solutions of even the hardest optimization problem, using differentiable programming, promises to provide us with entirely new and more performant or cheaper solutions to particle detection. I will look at the status of this research development and provide a few examples to prove its potential."

But to tell the truth, what I am most excited about is to participate to the conference dinner on Thursday evening, which will take place at the Musee d'Orsay. I visited the museum twice already in the past, but the place is so rich with incredible works of art that I am definitely looking forward to the tour before dinner...