USCMS Researcher: Di Croce Davide
Postdoc dates: Sep 2020 - Sep 2022
Home Institution: The University of Alabama
Project: Accelerating Machine Learning Reconstruction in CMS
Integration of innovative machine learning algorithms (end-to-end deep learning) into the CMS reconstruction code, leveraging heterogenous computing and hardware acceleration devices. Development of end-to-end (E2E) benchmarks, first integration of the end-to-end framework into CMS production code, and scaling studies of deep learning training and inference on heterogeneous computing devices. The new proposal extends this effort to novel E2E architectures, including graph neural networks, additional E2E reconstruction targets such as tau reconstruction and further E2E integration into CMS code.More information: My project proposal
Mentors:
-
Sergei Gleyzer (The University of Alabama)
- 19 May 2021 - "Accelerating End-to-End Deep Learning for Particle Reconstruction using CMS open data", Di Croce Davide, 25th International Conference on Computing in High-Energy and Nuclear Physics
Current Status
Postdoctoral Researcher at EPFL
Contact me: