# Adversarial Tuning of Perturbative Parameters in Non-Differentiable Physics Simulators

- Institution: CERN
- Department: Machine Learning
- Speaker: Michela Paganini
- Year: 2018
- Language: English
- Subjects: Computer science
- Source: [CERN](https://videos.cern.ch/record/3011350)
- 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)

In this contribution, we present a method for tuning perturbative parameters in Monte Carlo simulation using a classifier loss in high dimensions. We use an LSTM trained on the radiation pattern inside jets to learn the parameters of the final state shower in the Pythia Monte Carlo generator. This represents a step forward compared to unidimensional distributional template-matching methods..

0:00:00 Slide 1

0:00:18 Slide 2

0:00:29 Slide 3

0:00:34 Slide 4

0:01:13 Slide 5

0:02:48 Slide 6

0:04:49 Slide 7

0:05:13 Slide 8

0:06:15 Slide 9

0:07:50 Slide 10

0:08:10 Slide 11

0:08:26 Slide 12

0:10:48 Slide 13

0:13:07 Slide 14

0:13:59 Slide 15

0:14:22 Slide 16

0:15:40 Slide 17

0:16:01 Slide 18

0:16:16 Slide 19

0:17:18 Slide 20

0:17:53 Slide 21

0:19:34 Slide 22

0:19:39 Slide 23

0:19:44 Slide 24

0:19:48 Slide 25

0:19:54 Slide 26

0:20:05 Slide 27

0:20:10 Slide 28

0:20:17 Slide 29

0:20:24 Slide 30

0:20:38 Slide 31

0:21:14 Slide 32

0:21:42 Slide 33

0:21:47 Slide 34

0:21:51 Slide 35

0:22:25 Slide 36

0:23:13 Slide 37

## Lectures

1. [Adversarial Tuning of Perturbative Parameters in Non-Differentiable Physics Simulators](https://invisiblenetwork.ai/learn/cern/3011350/3e198c4bb137), 29 min (CERN's player)