Recurrent and Graph Neural Networks for Particle Tracking at the BM@N Experiment

Rusov, D.; Goncharov, Pavel; Shchavelev, E.; Lubchenkov, Leonid; Nikolskaya, A.; Rezvaya, E.; Ososkov, G.; Zhemchugov, A.

doi:10.1007/978-3-031-19032-2_32

Cited by 2 publications

(3 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The neural network designed to deal with highly contaminated data was proposed in [9], as the TrackNET neural model. It receives as input the coordinates of points of track candidates used as seeds and predicts the center and semi-axes of the ellipse on the next coordinate plane, on which the candidate track continuation is searched.…”

Section: Related Work and The Problem Formulationmentioning

confidence: 99%

“…There are two ways to eliminate such false tracks. The first one is proposed in [9] and consists of applying the combined approach, when the track candidates from the TrackNET output consider as an input of the graph neural network (GNN) described in [8]. A GNN can now observe the whole event but by looking only at hits from the potential track candidates, which are presented in the form of a graph.…”

Section: Related Work and The Problem Formulationmentioning

confidence: 99%

“…The training process took around 25.6 h to complete, with 500 epochs for training on a dataset comprising 100 noisy points and slightly more for a dataset comprising 1000 fake points. In Rusov et al [9], the authors present a combined approach based on TrackNetv3 and a GNN (graph neural network). The algorithm presented here in this paper describes a different approach to track elimination, where track candidates generated by the model are fed to a parallel track elimination algorithm with the same aim to reduce a false positive rate inherent to the TrackNetv2 model.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Algorithm for Enhancing Event Reconstruction Efficiency by Addressing False Track Filtering Issues in the SPD NICA Experiment

et al. 2023

View full text Add to dashboard Cite

This paper introduces methods for parallelizing the algorithm to enhance the efficiency of event recovery in Spin Physics Detector (SPD) experiments at the Nuclotron-based Ion Collider Facility (NICA). The problem of eliminating false tracks during the particle trajectory detection process remains a crucial challenge in overcoming performance bottlenecks in processing collider data generated in high volumes and at a fast pace. In this paper, we propose and show fast parallel false track elimination methods based on the introduced criterion of a clustering-based thresholding approach with a chi-squared quality-of-fit metric. The proposed strategy achieves a good trade-off between the effectiveness of track reconstruction and the pace of execution on today’s advanced multicore computers. To facilitate this, a quality benchmark for reconstruction is established, using the root mean square (rms) error of spiral and polynomial fitting for the datasets identified as the subsequent track candidate by the neural network. Choosing the right benchmark enables us to maintain the recall and precision indicators of the neural network track recognition performance at a level that is satisfactory to physicists, even though these metrics will inevitably decline as the data noise increases. Moreover, it has been possible to improve the processing speed of the complete program pipeline by 6 times through parallelization of the algorithm, achieving a rate of 2000 events per second, even when handling extremely noisy input data.

show abstract

Section: Related Work and The Problem Formulationmentioning

confidence: 99%

Section: Related Work and The Problem Formulationmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation