Working groups

The current pinnacle of AI are so-called foundation models (FMs) as pioneered in LLMs like ChatGPT or image generators like DALLE. Multimodal FMs centralize information from various data modalities and domains and encode them in a common meaningful latent representation. They offer an opportunity to go from narrow task-centric applications to general-purpose tasks. This WG will serve as a platform to discuss the potential transformative impact of FMs for fundamental science and discovery in particular, and how we can as a community foster progress.

Goals of the WG are: (1) Identify existing design paradigms for particle and astroparticle physics instruments which have become obsolete in the AI era, and assemble software strategies and research paths to overtake them. (2) Support the development of simulation tools that constitute enablers of co-design approaches to holistic optimization for detector use cases in HEP, astro-HEP, nuclear and neutrino physics. (3) Understand physical limits of information generated by particle interactions in granular calorimeters and conditions for its lossless extraction, as a preliminary step toward the AI-assisted hybridization of calorimeters and tracking detectors into optimized variable-density systems.

Tentative mission/goals: (1) Review existing activities exist in fundamental physics that strive to make ML algorithms Findable, Available, Interoperable and Reproducible (may include ML datasets. (2) Understand what the barriers to making ML FAIR are - this includes availability of standardised tools and computing facilities for retraining or using someone else’s model. (3) Discuss steps to facilitate and promote FAIR ML in fundamental science. (4) Start efforts to review and frame the environmental sustainability of ML algorithms (potential discussion of trade-off with performance)

The mission of the Joint ECFA-NuPECC-APPEC (JENA) Computing Initiative’s Machine Learning (ML) and Artificial Intelligence (AI) Working Group is to strategize and implement European federated computing solutions for future large-scale research facilities. Through a comprehensive survey, we aim to assess and quantify the resource needs of physicists to run ML workloads effectively.

ML in particle physics is a very small field, and the way we ususally operate is that we identify field-specific opportunities and challenges and then develop new solutions together with ML-experts from other fields (or based on their work). The obvious questions are (i) how can we make this communication more efficient? And, (ii) how can be organize a proper feedback loop, so our solutions are used by the ML community?