# Decoding Physics Information in DNNs

- Institution: CERN
- Department: LPCC Workshops
- Speaker: Taoli Cheng
- Year: 2019
- Language: English
- Subjects: Computer science
- Source: [CERN](https://videos.cern.ch/record/3016892)
- 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)

A more dedicated study on the information flow in DNNs will help us understand their behaviour and the deep connection between DNN models and the corresponding tasks. Taking into account our well-established physics analysis framework (observable-based), we present a novel way to interpret DNNs results for HEP, which not only gives a clear physics picture but also inspires interfaces with the theoretical foundation. Information captured by DNNs can thus be used as a fine-tailored general-purpose encoder. As a concrete example, we showcase using encoded information to help with physics searches at the LHC..

0:00:00 Slide 1

0:00:45 Slide 2

0:01:46 Slide 3

0:03:35 Slide 4

0:05:09 Slide 5

0:07:11 Slide 6

0:08:26 Slide 7

0:09:11 Slide 8

0:10:54 Slide 9

0:11:40 Slide 10

0:12:39 Slide 11

0:13:21 Slide 12

0:14:46 Slide 13

0:15:02 Slide 14

0:15:36 Slide 15

0:16:14 Slide 16

0:16:18 Slide 17

0:17:01 Slide 18

0:17:31 Slide 19

0:17:35 Slide 20

0:19:12 Slide 21

0:19:24 Slide 22

## Lectures

1. [Decoding Physics Information in DNNs](https://invisiblenetwork.ai/learn/cern/3016892/4c20f460ac19), 22 min (CERN's player)