POSTPONED to first half of February 2021
- Mon 20 - Wed 22 July : Workshop on Graph Neural Network
- Thu 23 - Sat 25 July morning : Workshop on Dealing with Uncertainties
- Mon 27 - Wed 29 July : Learning to Discover AI and Physics conference
- Thu 30 - Fri 31 July : Workshop on Generative Models
Learning To Discover is a program on Artificial Intelligence and High Energy Physics (HEP) to take place at Institut Pascal Paris-Saclay 20-31 July 2020, in its beautiful new building. Over the two weeks, three themes will be successively tackled during innovation-oriented sessions of 2-3 days each, with a three three days general conference on AI and physics in the middle.
Since 2014, the field of AI and HEP has grown exponentially. Physicists have realised quickly the potential of AI to deal with the large amount of complex data they are collecting and analysing. Many AI techniques have been put forward, with scientific collaboration based on open data sets, challenges, workshops and papers.
Learning To Discover is a first-of-its-kind program where participants will have access to deep technical insights in advanced machine learning techniques, and their application to particle physics. During the event, blending the concept of hackathon, hands-on and tutorials, physicists who have attempted to apply machine learning to specific challenges in HEP will expose their problem case and the solutions they have arrived at so far. ML experts will expose the latest advances in relevant techniques. Machine learning experts and ML-aware physicists will work hand in hand on existing datasets, building upon and improving existing solutions.Bibliography and open datasets will have been made available months earlier in order to make the event most productive. Under this particular environment, the participants will be able to discuss and understand shortcomings of existing solutions and develop novel architectures and methods to outperform on the specific problem.
Three themes have tentatively been selected based on the one hand, their interest for HEP, and the fact there is already a number of HEP teams working on it, on the other hand, their importance in the Machine Learning field.
Graph Neural Network (GNN) : despite promising applications to several HEP problems (tracking, identification, calorimeter reconstruction, particle flow, …) there remains several bottlenecks in the models being used. In particular the following topics are of interest: ways to extract graph-level information, mechanisms for information propagation over the graph, accurate graph generation, evolution of graph topology within the model, ...
Dealing with Uncertainties : physicists ultimately write papers with measurements which always include an assessment of uncertainties. How to evaluate the uncertainty on a trained model ? How to build confidence into a ML model and how to convince our peers about it ? How to deal with the uncertainties on the input of the models to maximise the overall accuracy ? Bayesian Networks, Adversarial architectures and other techniques are being explored.
Generative Models : High Energy Physics is well known for having built accurate simulators which are able to deliver simulation of real experiments, from the Quantum Field Theory equations dealing with the interactions of fundamental particles up to specific answers of electronic circuits. These simulations are accurate, but not perfect, and very slow. Proof of Concept of using Generative Models (GAN, VAE,...) are being developed but many issues are to be solved before they make it to production. How to reach sufficient accuracy over a large parameter space ? How to make sure multi-dimension correlations are well described ? How to deal with the irregularity of the detector (think of an image with pixels of different sizes and shapes) ?
Learning To Discover AI and Physics conference will deal with the most recent topics at the intersection of the two fields. It will propose a few keynote talks from leading scientists in the field as well as wide sample of on-going developments. Submission of abstracts is open.
Learning To Discover will be hosted at the new Université Paris-Saclay Institut Pascal, an ideal venue for such interactions to take place. Located in Orsay (close to Paris) in the heart of Université Paris-Saclay, it has offices for up to 60 people (40 the first week of the workshop), meeting rooms, amphitheaters (up to 120 seats) and all amenities for a successful workshop. In the usual Institut Pascal format, lunches and social programme is free. Accommodation can be booked in the area on Institut Pascal expense for a large fraction of the participants; travel is covered only if needed. See "Institut Pascal" item in the menu bar for more practical details.
A special issue of Computing and Software for the Big Science (a refereed journal) will be prepared with the most relevant contributions with a submission deadline 4 months after the workshop, leaving ample time for finalisation of the studies showcased at the workshop. A number of workshops concerning AI and Physics have been organised in recent years, notably “Deep Learning for the Physical Science” at NeurIPS 2017 (https://dl4physicalsciences.github.io/), “Machine Learning and the Physical Science” at Neurips 2019 (https://ml4physicalsciences.github.io) and “AI and physics” at AMLD 2020 in Lausanne (https://appliedmldays.org/tracks/ai-physics). Studies developed at “Learning To Discover” will have a high chance at being accepted at future such events.