# Fast Machine Learning at the Edge for HEP Experiments

- Institution: CERN
- Department: Detector Seminar
- Speaker: Sioni Summers
- Year: 2024
- Language: English
- Subjects: Computer science
- Source: [CERN](https://videos.cern.ch/record/3003895)
- Licence: [CERN](http://copyright.web.cern.ch)
- Who can watch: Tweens profiles and older (sources give no ratings; placed by Invisible Network from the source's own words)

Facing increasing data challenges at current and future HEP experiments, there is a growing demand for fast and efficient Machine Learning (ML) techniques deployed close to the detector. In this seminar, we explore the application of Fast ML in accelerating and optimizing ML algorithms for real-time analysis, triggering, and data reduction, focussing on two projects in use at CERN experiments: hls4ml and conifer. These tools offer a novel approach transforming ML models into efficient hardware descriptions suitable for implementation on field-programmable gate arrays (FPGAs) and other hardware accelerators. We present the state of the art techniques for compressing ML models for fast and lightweight inference. We highlight case studies and experimental results that have leveraged these tools and techniques to improve performance such as particle identification, event classification, and anomaly detection. Finally, we present key future development directions.

Coffee will be served at 10:30. .

0:00:00 Slide 1

0:01:33 Slide 2

0:02:49 Slide 3

0:05:13 Slide 4

0:07:30 Slide 5

0:10:43 Slide 6

0:12:08 Slide 7

0:14:06 Slide 8

0:16:59 Slide 9

0:19:06 Slide 10

0:20:23 Slide 11

0:22:34 Slide 12

0:25:29 Slide 13

0:27:29 Slide 14

0:29:28 Slide 15

0:31:52 Slide 16

0:33:12 Slide 17

0:35:30 Slide 18

0:37:16 Slide 19

0:39:07 Slide 20

0:41:05 Slide 21

0:42:27 Slide 22

0:43:58 Slide 23

0:45:13 Slide 24

0:46:31 Slide 25

0:48:37 Slide 26

0:49:44 Slide 27

0:51:36 Slide 28

0:53:38 Slide 29

0:54:57 Slide 30

0:56:17 Slide 31

0:57:47 Slide 32

1:00:50 Slide 33

1:01:55 Slide 34

## Lectures

1. [Fast Machine Learning at the Edge for HEP Experiments](https://invisiblenetwork.ai/learn/cern/3003895/ad544c8cf5a2), 1 h 7 min (CERN's player)