Reconstructing heavy particles from their observed decay products becomes complex when decay chains with many intermediate and final state particles are involved and requires solving ambiguities. Modern machine learning techniques offer new solutions to this task. We discuss new applications of machine-learning techniques for event-level particle reconstruction in CMS.
Machine learning has become an essential tool in jet physics. Due to their complex, high-dimensional nature, jets can be explored holistically by neural networks in ways that are not possible manually. However, innovations in all areas of jet physics are proceeding in parallel. We show that large machine learning models trained for a jet classification task can improve the accuracy,...
Multi-head attention based Transformers have taken the world by storm, given their outstanding capacity of learning accurate representations of diverse types of data. Famous examples include Large Language Models, such as ChatGPT, and Vision Transformers, like BEiT, for image generation. In this talk, we take these major technological advancements to the realm of jet physics. By creating a...
The search for heavy resonances beyond the Standard Model (BSM) is a key objective at the LHC. While the recent use of advanced deep neural networks for boosted-jet tagging significantly enhances the sensitivity of dedicated searches, it is limited to specific final states, leaving vast potential BSM phase space underexplored. In this talk, we introduce a novel experimental method,...
Attention-based transformer models have become increasingly prevalent in collider analysis, offering enhanced performance for tasks such as jet tagging. However, they are computationally intensive and require substantial data for training. In this paper, we introduce a new jet classification network using an MLP mixer, where two subsequent MLP operations serve to transform particle and feature...
The likelihood-ratio test can be used to perform a goodness-of-fit test between a reference model and observations if the alternative hypothesis is selected from data by exploring a rich parametrised family of functions. The New Physics Learning Machine (NPLM) methodology has been developed as a concrete realisation of this idea, to perform model-independent searches at collider experiments....
The Energy Mover’s Distance (EMD) has seen use in collider physics as a metric between events and as a geometric method of defining infrared and collinear safe observables. Recently, the spectral Energy Mover’s Distance (SEMD) has been proposed as a more analytically tractable alternative to the EMD. In this work, we obtain a closed-form expression for the Riemannian-like p = 2 SEMD metric...
It is known that perturbative simulations of high-multiplicity jets can vary quite substantially from data collected at high-energy colliders like the LHC. It is therefore important to understand what is driving the discrepancy and if there are other possible tools to simulate these events without relying on individual particle simulation. We propose that another observable to manage...
Energy correlators, which as a jet-substructure observable measure correlations between energy detectors (calorimeters) in a collider experiment, have received significant attention over the last few years in both the theory/phenomenology and experimental communities. This success has prompted investigations into how energy correlators can be further used, such as in the study of both hot and...
Abstract: The current best-performing networks in many ML for particle physics tasks are either custom-built Lorentz-equivariant architectures or more generic large transformer models. A major unanswered question is whether the high performance of equivariant architectures is in fact due to their equivariance. We design a study to isolate and investigate effects of equivariance on network...
We propose a new approach to learning powerful jet representations directly from unlabelled data. The method employs a Particle Transformer to predict masked particle representations in a latent space, overcoming the need for discrete tokenization and enabling it to extend to arbitrary input features beyond the Lorentz four-vectors. We demonstrate the effectiveness and flexibility of this...
How can one fully harness the power of physics encoded in relativistic $N$-body phase space? Topologically, phase space is isomorphic to the product space of a simplex and a hypersphere and can be equipped with explicit coordinates and a Riemannian metric. This natural structure that scaffolds the space on which all collider physics events live opens up new directions for machine learning...
