We present a novel application of the machine learning / artificial intelligence method called boosted decision trees to estimate physical quantities on field programmable gate arrays (FPGA). The software packagefwXmachinafeatures a new architecture called parallel decision paths that allows for deep decision trees with arbitrary number of input variables. It also features a new optimization scheme to use different numbers of bits for each input variable, which produces optimal physics results and ultraefficient FPGA resource utilization. Problems in high energy physics of proton collisions at the Large Hadron Collider (LHC) are considered. Estimation of missing transverse momentum (ETmiss) at the first level trigger system at the High Luminosity LHC (HL-LHC) experiments, with a simplified detector modeled by Delphes, is used to benchmark and characterize the firmware performance. The firmware implementation with a maximum depth of up to 10 using eight input variables of 16-bit precision gives a latency value of $$\mathcal{O}$$(10) ns, independent of the clock speed, and $$\mathcal{O}$$(0.1)% of the available FPGA resources without using digital signal processors.
Carlson, B. T., et al. "Nanosecond machine learning regression with deep boosted decision trees in FPGA for high energy physics." Journal of Instrumentation, vol. 17, no. 09, Sep. 2022. https://doi.org/10.1088/1748-0221/17/09/p09039
Carlson, B. T., Bayer, Q., Hong, T. M., & Roche, S. T. (2022). Nanosecond machine learning regression with deep boosted decision trees in FPGA for high energy physics. Journal of Instrumentation, 17(09). https://doi.org/10.1088/1748-0221/17/09/p09039
Carlson, B. T., Bayer, Q., Hong, T. M., et al., "Nanosecond machine learning regression with deep boosted decision trees in FPGA for high energy physics," Journal of Instrumentation 17, no. 09 (2022), https://doi.org/10.1088/1748-0221/17/09/p09039
@article{osti_1979442,
author = {Carlson, B. T. and Bayer, Q. and Hong, T. M. and Roche, S. T.},
title = {Nanosecond machine learning regression with deep boosted decision trees in FPGA for high energy physics},
annote = {We present a novel application of the machine learning / artificial intelligence method called boosted decision trees to estimate physical quantities on field programmable gate arrays (FPGA). The software packagefwXmachinafeatures a new architecture called parallel decision paths that allows for deep decision trees with arbitrary number of input variables. It also features a new optimization scheme to use different numbers of bits for each input variable, which produces optimal physics results and ultraefficient FPGA resource utilization. Problems in high energy physics of proton collisions at the Large Hadron Collider (LHC) are considered. Estimation of missing transverse momentum (ETmiss) at the first level trigger system at the High Luminosity LHC (HL-LHC) experiments, with a simplified detector modeled by Delphes, is used to benchmark and characterize the firmware performance. The firmware implementation with a maximum depth of up to 10 using eight input variables of 16-bit precision gives a latency value of $\mathcal{O}$(10) ns, independent of the clock speed, and $\mathcal{O}$(0.1)% of the available FPGA resources without using digital signal processors.},
doi = {10.1088/1748-0221/17/09/p09039},
url = {https://www.osti.gov/biblio/1979442},
journal = {Journal of Instrumentation},
issn = {ISSN 1748-0221},
number = {09},
volume = {17},
place = {United States},
publisher = {Institute of Physics (IOP)},
year = {2022},
month = {09}}
Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, Vol. 568, Issue 2https://doi.org/10.1016/j.nima.2006.07.053