Test

FastML Lab

Real-time and accelerated ML for fundamental sciences

Fast ML Lab is a research collective of physicists, engineers, and computer scientists interested in deploying machine learning algorithms for unique and challenging scientific applications. Our projects range from real-time, on-detector and low latency machine learning applications to high-throughput heterogeneous computing big data challenges. We are interested in deploying sophisticated machine learning algorithms to advance the exploration of fundamental physics from the world’s biggest colliders to the most intense particle beams to the cosmos.

News

It’s official! The registration and call for abstracts for the 2022 workshop is now open. It will be a hybrid workshop hosted by Southern Methodist University from Oct 3 – Oct 6, 2022. To see past editions, visit our virtual 2020 workshop and 2019 workshop webpages.

CERN Courier

CERN Courier

An article in the CERN Courier about AI Triggers at CERN titled "Hunting anomalies with an AI trigger"

Symmetry Magazine

Symmetry Magazine

Symmetry Magazine article about fast AI trigger systems in particle physics titled "Blink and it's gone"

Tiny MLPerf

Tiny MLPerf

hls4ml team submits results for the first Tiny MLPerf Inference Benchmark

Nvidia Triton

Nvidia Triton

Scaling Inference in High Energy Particle Physics at Fermilab Using NVIDIA Triton Inference Server

HLS4ML

HLS4ML

hls4ml: an open-source code framework for translating machine learning algorithms directly into FPGA firmware

Zenuity

Zenuity

Zenuity has become the first automotive company to team up with CERN to develop ML for autonomous drive cars

Microsoft Azure

Microsoft Azure

Microsoft Azure collaboration deploying FPGAs to accelerate ML to prototype computing solutions for future big science experiments

Xilinx Case Study

Xilinx Case Study

A story from Xilinx on how we use high level synthesis to find the best events at the Large Hadron Collider

Papers

Real-time Artificial Intelligence for Accelerator Control: A Study at the Fermilab Booster, arXiv:2011.07371.
A reconfigurable neural network ASIC for detector front-end data compression at the HL-LHC, IEEE Trans. Nucl. Sci. 68, 2179 (2021).
Autoencoders on FPGAs for real-time, unsupervised new physics detection at 40 MHz at the Large Hadron Collider, arXiv:2108.03986.
Accelerating Recurrent Neural Networks for Gravitational Wave Experiments, ASAP Conference 2021.
Fast convolutional neural networks on FPGAs with hls4ml, Mach. Learn.: Sci. Technol. 2, 045015 (2021).
MLPerf Tiny Benchmark, NeurIPS 2021 Datasets and Benchmarks Track.
hls4ml: An Open-Source Codesign Workflow to Empower Scientific Low-Power Machine Learning Devices, tinyML Reserach Symposium 2021.
GPU-accelerated machine learning inference as a service for computing in neutrino experiments, arXiv:2009.04509 [physics.comp-ph].
Distance-Weighted Graph Neural Networks on FPGAs for Real-Time Particle Reconstruction in High Energy Physics, arXiv:2008.03601 [physics.comp-ph].
GPU coprocessors as a service for deep learning inference in high energy physics, arXiv:2007.10539 [physics.comp-ph].
Ultra Low-latency, Low-area Inference Accelerators using Heterogeneous Deep Quantization with QKeras and hls4ml, arXiv:2006.10159 [physics.ins-det].
Compressing deep neural networks on FPGAs to binary and ternary precision with hls4ml, MLST (2020).
Fast inference of Boosted Decision Trees in FPGAs for particle physics, JINST 15, P05026 (2020).
ESP4ML: Platform-Based Design of Systems-on-Chip for Embedded Machine Learning, DATE Conference 2020 .
Accelerated Machine Learning as a Service for Particle Physics Computing, NeurIPS ML4PS Workshop 2019.
Low-latency machine learning inference on FPGAs, NeurIPS ML4PS Workshop 2019.
Fast inference of deep neural networks in FPGAs for particle physics, JINST 13, P07027 (2018).
FPGA-accelerated machine learning inference as a service for particle physics computing, Comput. Softw. Big Sci. 3, 13 (2019)

Talks

C. Herwig, An ML Control System for the Fermilab Booster, BIDS Machine Learning and Science Forum, April 2021, abstract
P. Harris, Quick and Quirk with Quarks, IAIFI Colloquium Online, March 2021, video
P. Harris, Scientific Applications of FPGAs at the LHC, ISFPGA 2021 (keynote), abstract
J. Duarte, AI at the Edge of Particle Physics, video
J. Duarte, hls4ml: An open-source codesign workflow to empower scientific low-power machine learning devices, tinyML Research Symposium 2021, video
D. Rankin, FPGAs-as-a-Service Toolkit (FaaST), Heterogeneous High-Performance Reconfigurable Computing Workshop at Supercomputing 2020, slides
P. Harris, ML Acceleration with Heterogeneous Computing for Big Data Physics Experiments, Heterogeneous High Performance Computing Workshop at Supercomputing 2019, slides
K. Pedro, FPGA-accelerated machine learning inference as a service for particle physics computing, CHEP 2019, slides
J. Duarte, Machine Learning on FPGAs for low latency and high throughput inference, eScience 2019, slides
M. Liu, FPGA-accelerated machine learning inference as a solution for particle physics computing challenges, PASC 2019, slides
J. Ngadiuba, hls4ml: deploying deep learning on FPGAs for trigger and data acquisition, ACAT 2019, slides
J. Duarte, FPGA-accelerated machine learning inference for particle physics computing challenges, CTD 2019, slides
J. Duarte, hls4ml: deploying deep learning on FPGAs for L1 trigger and data acquisition, TWEPP 2018, slides
J. Ngadiuba, Synthesizing machine learning algorithms on FPGAs, CHEP 2018, slides
N. Tran, Neural networks in FPGAs for trigger and DAQ, CTD 2018, slides

People

CALTECH

Jennifer Ngadiuba (PhD, Physics);

CERN

Thea Årrestad (PhD, Physics); Vladimir Loncar (PhD, Computer Science); Maurizio Pierini (PhD, Physics); Sioni Summers (PhD, Physics);

COLUMBIA UNIVERSITY

Giuseppe Di Guglielmo (PhD, Computer Science);

FERMILAB

Lindsey Gray (PhD, Physics); Christian Herwig (PhD, Physics); Burt Holzman (PhD, Physics); Sergo Jindariani (PhD, Physics); Thomas Klijnsma (PhD, Physics); Ben Kreis (PhD, Physics); Kevin Pedro (PhD, Physics); Ryan Rivera (PhD, EE); Nhan Tran (PhD, Physics); Mike Wang (PhD, Physics); Tingjun Yang (PhD, Physics);

HAWKEYE 360

EJ Kreinar ( Computer Science);

LEHIGH UNIVERSITY

Joshua Agar (PhD, Material Science and Engineering);

MIT

Jack Dinsmore (Undergraduate, Physics); Song Han (PhD, EECS); Phil Harris (PhD, Physics); Duc Hoang (Graduate, Physics); Jeffrey Krupa (Graduate, Physics); Eric Moreno (Graduate, Physics); Noah Paladino (Graduate, Physics); Sang Eon Park (Graduate, Physics); Dylan Rankin (PhD, Physics);

NORTHWESTERN UNIVERISITY

Farah Fahim (Adjunct, ECE); Seda Memik-Ogrenci ( ECE); Nhan Tran (Adjunct, ECE);

PURDUE UNIVERSITY

Mia Liu (PhD, Physics);

UC SAN DIEGO

Daniel Diaz (PhD, Physics); Javier Duarte (PhD, Physics); Ryan Kastner (PhD, Computer Science); Farouk Mokhtar (Graduate, Physics); Tai Nguyen (Undergraduate, Physics); Vesal Razavimaleki (Undergraduate, Engineering Physics); Rushil Roy (Undergraduate, Electrical Engineering); Olivia Weng (Graduate, Computer Science and Engineering);

UNIVERSITY OF ILLINOIS CHICAGO

Zhenbin Wu (PhD, Physics);

UNIVERSITY OF ILLINOIS URBANA-CHAMPAIGN

Markus Atkinson (PhD, Physics); Mark Neubauer (PhD, Physics);

UNIVERSITY OF WASHINGTON

Scott Hauck (PhD, EECS); Shih-Chieh Hsu (PhD, Physics);

Collaboration

We are a community that fosters knowledge transfer of accelerated and real-time artificial intelligence applications to fundamental science. By joining the community, you agree to abide by our codes of conduct and policies of collaboration.

Code of Conduct

Sponsors and Partners

We are graciously funded by the following organizations:

European Research Council

Department of Energy

National Science Foundation

Internet 2

DARPA

And are collaborating with these organizations:

Institute for Research and Innovation in Software for High Energy Physics

Institute for Accelerated AI Algorithms for Data-Driven Discovery