arXiv subjects

Tristan Brandes

Publications and source records attributed to Tristan Brandes.

3 recordsLinked to original sources

arXiv2026

Mitigating Detector Ageing Effects with Graph-Based Multi-Modal Track Reconstruction at Belle II

Large backgrounds that can lead to hardware failures and the degradation of detector gain impact the track finding in the Belle II central drift chamber. These conditions lead to spatially non-uniform and time-dependent inefficiencies, which results in inactive regions and missing hits which challenges conventional tracking algorithms and necessitate the development of new track finding algorithms. In this work, we evaluate the performance of our previously developed unified graph neural network (GNN) based track-finding algorithm under realistic long-term detector ageing conditions. Track finding is formulated as a global relational clustering problem using object condensation, which enables the reconstruction of an unknown and variable number of tracks per event. Using a realistic full detector simulation incorporating beam-induced backgrounds, detector noise, and measured detector ageing effects, we evaluate the tracking performance and compare it to the current Belle II baseline reconstruction. We show that detector degradation can be treated as a domain shift in the observed hit patterns, rather than requiring a fundamentally new reconstruction strategy. After retraining on degraded detector conditions, the GNN-based approach limits the absolute track efficiency loss for uniformly displaced muons to 14%, compared to 28% for the baseline tracking, while maintaining a track purity of 96%. Under the same conditions, the baseline reconstruction achieves only 90% track purity. These results demonstrate that the unified GNN-based reconstruction provides increased robustness to irregular hit patterns and extended inactive regions, enabling stable tracking performance under long-term detector ageing at Belle II.

hep-ex↗

arXiv2026

Multi-Modal Track Reconstruction using Graph Neural Networks at Belle II

High backgrounds and detector ageing impact the track finding in the Belle II central drift chamber, reducing both track purity and track efficiency in events. This necessitates the development of new track finding algorithms to mitigate detector performance degradation. Building on our previous success with an end-to-end multi-track reconstruction algorithm for the Belle II experiment at the SuperKEKB collider (arXiv:2411.13596), we have extended the algorithm to incorporate inputs from both the drift chamber and the silicon vertex tracking detector, creating a multi-modal network. We employ graph neural networks to handle the irregular detector structure and object condensation to address the unknown, varying number of particles in each event. This approach simultaneously identifies all tracks in an event and determines their respective parameters. We demonstrate the algorithm's effectiveness using a realistic full detector simulation, which incorporates beam-induced backgrounds and noise modelled from actual collision data. The simultaneous reconstruction of the information from the two detectors yields a track efficiency improvement from 48.0 % to 74.7 % for uniformly displaced particles up to 100 cm, while increasing the track purity by 5.5 percentage points. We provide a detailed comparison of its track-finding performance against the current Belle II baseline across various event topologies.

hep-ex↗

arXiv2024

Distilling particle knowledge for fast reconstruction at high-energy physics experiments

Knowledge distillation is a form of model compression that allows artificial neural networks of different sizes to learn from one another. Its main application is the compactification of large deep neural networks to free up computational resources, in particular on edge devices. In this article, we consider proton-proton collisions at the High-Luminosity LHC (HL-LHC) and demonstrate a successful knowledge transfer from an event-level graph neural network (GNN) to a particle-level small deep neural network (DNN). Our algorithm, DistillNet, is a DNN that is trained to learn about the provenance of particles, as provided by the soft labels that are the GNN outputs, to predict whether or not a particle originates from the primary interaction vertex. The results indicate that for this problem, which is one of the main challenges at the HL-LHC, there is minimal loss during the transfer of knowledge to the small student network, while improving significantly the computational resource needs compared to the teacher. This is demonstrated for the distilled student network on a CPU, as well as for a quantized and pruned student network deployed on a field-programmable gate array. Our study proves that knowledge transfer between networks of different complexity can be used for fast artificial intelligence (AI) in high-energy physics that improves the expressiveness of observables over non-AI-based reconstruction algorithms. Such an approach can become essential at the HL-LHC experiments, e.g., to comply with the resource budget of their trigger stages.

hep-ex↗