IRIS-HEP Fellow: Nishank Gite

Fellowship dates: Jun – Sep, 2024

Home Institution: University of California, Berkeley

Project: Towards Differentiable Jet Clustering

Our research is centered on the differentiation of jet clustering methods, which presents a key challenge in existing algorithms like anti-kt and the more recent qanti-kt. These conventional approaches yield distinct results, making them non-differentiable and creating significant hurdles for optimization through neural networks, especially in forward automatic differentiation and backpropagation. Our goals are to address this issue by converting discrete and deterministic clustering into a probabilistic framework through the use of expectations. We also aim to employ this framework in the context of neural networks for initial attempts at jet clustering optimization.

More information: My project proposal

Mentors:

• Nicole Hartman (TUM)

• Annalena Kofler (MPI-IS)

• Michael Kagan (SLAC)

• Lukas Heinrich (TUM)

Presentations and Publications

• 16 Oct 2024 - "Differentiable Jet Clustering", Nishank Gite, IRIS-HEP Fellows Presentations 2024 Recording: Differentiable Jet Clustering

---

Current Status


Contact me:

• Send me an email
  
• Visit my GitHub Repo
  
• Visit my LinkedIn Profile