Related papers: Towards Foundation Models for Experimental Readout…

The physics potential of high intensity conventional beams is explored. We consider a low energy super beam which could be produced by a proposed new accelerator at CERN, the Super Proton Linac. Water Cherenkov and liquid oil scintillator…

A new computational model for the description of the photon detector response functions measured in conditions of low light is presented, together with examples of the observed photomultiplier signal amplitude distributions, successfully…

Foundation models, often pre-trained with large-scale data, have achieved paramount success in jump-starting various vision and language applications. Recent advances further enable adapting foundation models in downstream tasks efficiently…

Water Cherenkov detectors offer a robust and economical solution for real-time radiation monitoring by detecting Cherenkov light from charged particles moving faster than light in water. This work presents a novel two-stage classification…

We suggest and implement an approach for the bottom-up description of systems undergoing large-scale structural changes and chemical transformations from dynamic atomically resolved imaging data, where only partial or uncertain data on…

This paper describes the construction of novel end-to-end image-based classifiers that directly leverage low-level simulated detector data to discriminate signal and background processes in pp collision events at the Large Hadron Collider…

Genomic foundation models have the potential to decode DNA syntax, yet face a fundamental tradeoff in their input representation. Standard fixed-vocabulary tokenizers fragment biologically meaningful motifs such as codons and regulatory…

Autoregressive (AR) language models rely on causal tokenization, but extending this paradigm to vision remains non-trivial. Current visual tokenizers either flatten 2D patches into non-causal sequences or enforce heuristic orderings that…

Conditional discrete generative models struggle to faithfully compose multiple input conditions. To address this, we derive a theoretically-grounded formulation for composing discrete probabilistic generative processes, with masked…

Conditional discrete generative models struggle to faithfully compose multiple input conditions. To address this, we derive a theoretically-grounded formulation for composing discrete probabilistic generative processes, with masked…

Tabular foundation models like TabPFN and TabICL achieve state-of-the-art performance through in-context learning, yet their architectures remain fundamentally opaque. We introduce KernelICL, a framework to enhance tabular foundation models…

Navigating topological transitions in cellular mechanical systems is a significant challenge for existing simulation methods. While abstract models lack predictive capabilities at the cellular level, explicit network representations…

In this work we present a new method of approximating the continuum wavefunctions with a discrete basis set. This method aims to be at least compatible with other well known methods of the electronic structure theory to describe processes…

In recent years, test-time adaptive object detection has attracted increasing attention due to its unique advantages in online domain adaptation, which aligns more closely with real-world application scenarios. However, existing approaches…

Foundational Machine Learning Potentials can resolve the accuracy and transferability limitations of classical force fields. They enable microscopic insights into material behavior through Molecular Dynamics simulations, which can crucially…

We introduce a hierarchical framework that combines ab initio many-body calculations with a Bayesian neural network, developing emulators capable of accurately predicting nuclear properties across isotopic chains simultaneously and being…

Advances in Earth observation (EO) foundation models have unlocked the potential of big satellite data to learn generic representations from space, benefiting a wide range of downstream applications crucial to our planet. However, most…

The current focus of the CERN program is the Large Hadron Collider (LHC), however, CERN is engaged in long baseline neutrino physics with the CNGS project and supports T2K as recognized CERN RE13, and for good reasons: a number of observed…

We discuss a method that employs a multilayer perceptron to detect deviations from a reference model in large multivariate datasets. Our data analysis strategy does not rely on any prior assumption on the nature of the deviation. It is…

This paper presents Diffusion Forcing, a new training paradigm where a diffusion model is trained to denoise a set of tokens with independent per-token noise levels. We apply Diffusion Forcing to sequence generative modeling by training a…