Ultrafast jet classification at the HL-LHC

← Back to all benchmarks

Date: 2024-07-08

Name: Ultrafast jet classification at the HL-LHC

Domain: High Energy Physics

Focus: FPGA-optimized real-time jet origin classification at the HL-LHC

Task Types: Classification

Metrics: Accuracy, Latency, Resource utilization

Models: MLP, Deep Sets, Interaction Network

AI/ML Motif: Classification

Resources

Benchmark: Visit

Datasets: Zenodo dataset

Results: Gemini LLM Deep Research, ChatGPT LLM

Keywords

jet classification FPGA quantization-aware training Deep Sets Interaction Networks

Citation

• Patrick Odagiu, Zhiqiang Que, Javier Duarte, Johannes Haller, Gregor Kasieczka, Artur Lobanov, Vladimir Loncar, Wayne Luk, Jennifer Ngadiuba, Maurizio Pierini, Philipp Rincke, Arpita Seksaria, Sioni Summers, Andre Sznajder, Alexander Tapper, and Thea K. Aarrestad. Ultrafast jet classification on fpgas for the hl-lhc. 2024. URL: https://arxiv.org/abs/2402.01876, arXiv:2402.01876, doi:https://doi.org/10.1088/2632-2153/ad5f10.

@misc{odagiu2024ultrafastjetclassificationfpgas,
  archiveprefix = {arXiv},
  author        = {Patrick Odagiu and Zhiqiang Que and Javier Duarte and Johannes Haller and Gregor Kasieczka and Artur Lobanov and Vladimir Loncar and Wayne Luk and Jennifer Ngadiuba and Maurizio Pierini and Philipp Rincke and Arpita Seksaria and Sioni Summers and Andre Sznajder and Alexander Tapper and Thea K. Aarrestad},
  doi           = {https://doi.org/10.1088/2632-2153/ad5f10},
  eprint        = {2402.01876},
  primaryclass  = {hep-ex},
  title         = {Ultrafast jet classification on FPGAs for the HL-LHC},
  url           = {https://arxiv.org/abs/2402.01876},
  year          = {2024}
}

Ratings

CategoryRating

Software

3.00

Not containerized; Setup and automation incomplete

Specification

4.00

Hardware constraints are referenced but not fully detailed or standardized

Dataset

4.00

FAIR metadata limited; no clear mention of dataset format or splits

Metrics

3.00

Metrics exist (accuracy, latency, utilization), but formal definitions and evaluation guidance are limited

Reference Solution

2.00

Reference implementations not fully reproducible; no evaluation pipeline or training setup provided

Documentation

3.00

No linked GitHub repo or setup instructions; paper provides partial guidance only

Average rating: 3.17/5

Radar plot

Edit: edit this entry