← Back to all news
arXiv AI By Max Marriott-Clarke, Lazar Novakovic, Elizabeth Ratzer, Robert J. Bainbridge, Loukas Gouskos, Benedikt Maier

Contrastive Metric Learning for Point Cloud Segmentation in Highly Granular Detectors

Read the original on arXiv AI →

The paper introduces a clustering method for point‑cloud segmentation that uses supervised contrastive metric learning (CML) to embed points from the same particle shower close together while separating unrelated points. After learning this latent space, clusters are reconstructed with a density‑based readout, decoupling representation learning from cluster formation. Compared to object condensation, CML yields a more stable, separable embedding, improving local neighbourhood consistency, shower separation, and generalization to unseen multiplicities and energies, which translates into higher reconstruction efficiency, purity, and energy resolution in highly granular calorimeter data.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Related stories

arXiv Machine Learning
Sep 14

Learning the Geometry of Collider Events with Metric-Aware Deep Sets

The paper introduces a Deep Sets surrogate for optimal transport (OT) that respects key metric properties—non-negativity, exchange symmetry, and zero self-distance—while leaving the triangle inequality unconstrained. Applied to the Energy Mover's Distance between collider events, the Metric-Aware Particle Flow Network achieves percent‑level mean absolute percentage error and markedly higher inference throughput compared to other exact and approximate methods. The architectural constraints also dramatically reduce triangle‑inequality violations, improving geometric fidelity across a large set of held‑out event triplets.

By Lauren Hay, Rishabh Jain, Matt LeBlanc, Jennifer Roloff
efficiency
More like this →
arXiv Computer Vision
Sep 2

Panda Diplomacy: Foundation Model Pre-training across Particle Imaging Detectors for High Energy and Nuclear Physics

Panda Diplomacy introduces a point‑cloud self‑distillation framework that enables a single foundation‑model architecture and objective to be pre‑trained across three distinct particle‑detector modalities—liquid argon time‑projection chambers, collider TPCs, and water Cherenkov detectors—without extensive modification. Using only 1,000 labeled images for downstream adaptation, the resulting Panda V2 model matches or surpasses specialized baselines that require orders of magnitude more supervision, achieving state‑of‑the‑art particle‑clustering performance with 70× fewer labeled events on sPHENIX and up to 1,000× fewer labels on LArTPC data. Linear probes further demonstrate that the model’s latent space captures physically meaningful structures such as particle causality and track curvature.

By Samuel Young, C\'esar Jes\'us-Valls, Kazuhiro Terao
efficiencybenchmarks
More like this →