IAIFI 2025 Summer School & Workshop — Harvard University

A Long Overdue Post

This summer, I had the incredible opportunity to participate in the NSF AI Institute for Artificial Intelligence and Fundamental Interactions (IAIFI) Summer School and Workshop at Harvard University — an experience that was as intense as it was inspiring.

Summer School

The summer school (August 4–8, 2025) was a deep dive into cutting-edge topics at the intersection of AI and physics, with lectures from world-class researchers:

• Sasha Rakhlin — Reinforcement Learning
• Aleksandra Ciprijanović — Domain Shift & Building Robust AI
• Gaia Grosso — Physics-Motivated Optimization
• SueYeon Chung — Representation Manifold Learning & Geometric Deep Learning

The days were packed with tutorials, discussions, and collaborative problem-solving. But beyond the lectures, the real magic happened in the conversations between researchers — exchanging ideas, challenging assumptions, and finding unexpected intersections between disciplines.

Contributed Talk — Workshop

During the Summer Workshop, I presented my work on “Data-Driven Classification of Structural Nonlinearities Using Interpretable Deep Learning on Time Series” — part of the research behind the Benchmarking Structural Nonlinearities with Interpretable Deep Learning project.

One fascinating takeaway was seeing how interpretability is understood differently across domains:

• In structural dynamics, interpretability focuses on ante-hoc or post-hoc explainable AI methods applied to sensor time series.
• For many physicists, it means interpreting model results or controlling AI variables directly within a physical simulation.

That clash of perspectives sparked some of the most thought-provoking conversations I had all year.

People & Places

Equally memorable were the people. It was wonderful to connect with Gouda, Zhongtian Hu, Jake, Kaitlin, and Mohamed Nagy as we explored Boston together and discussed our research. I was also glad to reconnect with my MathEXLab team member Keane Ong, who graciously showed us around MIT.

I’m deeply grateful to NSF-IAIFI for this opportunity and excited to integrate these insights into my ongoing research. Experiences like this remind me why we do what we do — to push boundaries, challenge paradigms, and bridge disciplines.

#AI #DeepLearning #Physics #ReinforcementLearning #InterpretableAI #Research #NSFIAIFI #Harvard #MIT