Boosted Decision Trees for CMS Muon Triggers

This episode explores how the CMS experiment uses machine learning inside its Level-1 endcap muon trigger, where hardware must estimate muon momentum within roughly 500 nanoseconds while filtering an enormous stream of proton-collision data. It explains why boosted decision trees were chosen over neural networks: not because they are trendier, but because they fit strict FPGA constraints around deterministic latency, fixed-point arithmetic, and bounded memory. A central finding is that the online system does not run the trees directly; instead, the model is trained offline and compiled into a massive precomputed lookup table, turning inference into a single fast memory access. The discussion is especially interesting because it shows machine learning as a systems-and-hardware co-design problem, grounded in detector physics, feature engineering, and the practical realities of deploying learned functions in one of the harshest real-time environments in science. Sources: 1. Boosted Decision Trees for CMS Muon Triggers https://indico.cern.ch/event/567550/papers/2629686/files/6172-acat_bdt_l1t.pdf 2. Applications and Techniques for Fast Machine Learning in Science — Allison McCarn Deiana, Nhan Tran, Joshua Agar, Michaela Blott, Giuseppe Di Guglielmo, Javier Duarte, Philip Harris, Mia Liu, Mark Neubauer, Jennifer Ngadiuba, Maurizio Pierini and many others, 2022 https://scholar.google.com/scholar?q=Applications+and+Techniques+for+Fast+Machine+Learning+in+Science 3. Fast inference of Boosted Decision Trees in FPGAs for particle physics — Sioni Summers, Giuseppe Di Guglielmo, Javier Duarte, Philip Harris, Duc Hoang, Sergo Jindariani, Edward Kreinar, Vladimir Loncar, Jennifer Ngadiuba, Maurizio Pierini, Dylan Rankin, Nhan Tran and Zhenbin Wu, 2020 https://scholar.google.com/scholar?q=Fast+inference+of+Boosted+Decision+Trees+in+FPGAs+for+particle+physics 4. Automatic heterogeneous quantization of deep neural networks for low-latency inference on the edge for particle detectors — Claudionor N. Coelho Jr, Aki Kuusela, Shan Li, Hao Zhuang, Jennifer Ngadiuba, Thea Klaeboe Aarrestad, Vladimir Loncar, Maurizio Pierini, Adrian Alan Pol and Sioni Summers, 2021 https://scholar.google.com/scholar?q=Automatic+heterogeneous+quantization+of+deep+neural+networks+for+low-latency+inference+on+the+edge+for+particle+detectors 5. Serving DNNs in Real Time at Datacenter Scale with Project Brainwave — Eric Chung, Jeremy Fowers, Kalin Ovtcharov, Michael Papamichael, Adrian Caulfield, Todd Massengill, Ming Liu, Mahdi Ghandi, Daniel Lo and others, 2018 https://scholar.google.com/scholar?q=Serving+DNNs+in+Real+Time+at+Datacenter+Scale+with+Project+Brainwave 6. The CMS Trigger System — CMS Collaboration, not specified in excerpt https://scholar.google.com/scholar?q=The+CMS+Trigger+System 7. The CMS Endcap Muon Track Finder — CMS Collaboration or EMTF-related authors, not specified in excerpt https://scholar.google.com/scholar?q=The+CMS+Endcap+Muon+Track+Finder 8. TMVA: Toolkit for Multivariate Data Analysis — Andreas Hoecker and collaborators, not specified in excerpt https://scholar.google.com/scholar?q=TMVA%3A+Toolkit+for+Multivariate+Data+Analysis 9. Fast Machine Learning for Science: how accelerated hardware and software are enabling real-time data analysis at the edge — Javier Duarte and collaborators, 2022 https://scholar.google.com/scholar?q=Fast+Machine+Learning+for+Science%3A+how+accelerated+hardware+and+software+are+enabling+real-time+data+analysis+at+the+edge 10. hls4ml: An Open-Source Codesign Workflow to Empower Scientific Low-Power Machine Learning Devices — Giuseppe Di Guglielmo, Javier Duarte and collaborators, 2021 https://scholar.google.com/scholar?q=hls4ml%3A+An+Open-Source+Codesign+Workflow+to+Empower+Scientific+Low-Power+Machine+Learning+Devices 11. End-to-end codesign of Hessian-aware quantized neural networks for FPGAs and ASICs — Javier Campos, Zhen Dong, Javier Duarte, Nhan Tran, et al., 2023 https://scholar.google.com/scholar?q=End-to-end+codesign+of+Hessian-aware+quantized+neural+networks+for+FPGAs+and+ASICs 12. FPGA-QNN: Quantized Neural Network Hardware Acceleration on FPGAs — Mustafa Tasci, Ayhan Istanbullu, Vedat Tumen, Selahattin Kosunalp, 2025 https://scholar.google.com/scholar?q=FPGA-QNN%3A+Quantized+Neural+Network+Hardware+Acceleration+on+FPGAs 13. An FPGA-Based Time-to-Digital Converter with Online Dual-Chain Calibration — Zhengsen Jia, Yuzhuo Wang, Jie Ding, Qian Xu, et al., 2025 https://scholar.google.com/scholar?q=An+FPGA-Based+Time-to-Digital+Converter+with+Online+Dual-Chain+Calibration 14. A Novel FPGA-based Time-to-Digital Converter featuring Machine Learning-Aided Self-Calibration — Arash Amini Bardpareh, Eleonora Vacca, Davide Nicolini, Luca Sterpone, et al., 2026 https://scholar.google.com/scholar?q=A+Novel+FPGA-based+Time-to-Digital+Converter+featuring+Machine+Learning-Aided+Self-Calibration 15. AI Post Transformers: FlatAttention for Tile-Based Accelerator Inference — Hal Turing & Dr. Ada Shannon, 2026 https://podcast.do-not-panic.com/episodes/2026-04-04-flatattention-for-tile-based-accelerator-56e6ca.mp3 16. AI Post Transformers: SolidAttention: Co-Designing Sparse Attention and SSD I/O — Hal Turing & Dr. Ada Shannon, 2026 https://podcast.do-not-panic.com/episodes/2026-03-18-solidattention-co-designing-sparse-atten-5a8622.mp3 Interactive Visualization: Boosted Decision Trees for CMS Muon Triggers