BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Summer school "Deep Learning and applications"
DTSTART;VALUE=DATE-TIME:20220829T070000Z
DTEND;VALUE=DATE-TIME:20220902T140000Z
DTSTAMP;VALUE=DATE-TIME:20220810T090239Z
UID:indico-event-8012@indico.math.cnrs.fr
DESCRIPTION:Deep learning is becoming a widely used tool in science. This
summer school is dedicated to different mathematical aspects of neural net
works\, with a special focus on applications in computer science and astro
physics.\n\nThe summer school is composed of five lectures and three talks
. Each lectures (2x1h30) will be complemented by 2 hours of tutorials.\n\n
It is primarily intended for the students of the graduate program "Mathema
tics and interactions : research and interactions" of the University of St
rasbourg but is also accessible to any interested PhD student or researche
r. No machine learning background is required to attend the summer school.
This event is supported by the Interdisciplinary Thematic Institute IRMIA
++.\n\nIt will take place from 29 August to 2 September in the IRMA confer
ence room at the University of Strasbourg.\n\n \n\n\nLectures\n\n\n Intro
duction to Deep Learning\, Léo Bois (Université de Strasbourg)\n Convolu
tional Neural Networks for object dectection: fast and accurate results wi
th the YOLO (You Only Look Once) method\, David Cornu (Observatoire de Par
is)\n\n\nThe objective of this series of courses is to provide a theoretic
al and practical overview of the YOLO (You Only Look Once) object detectio
n method. The first session will introduce the different Convolutional Neu
ral Network (CNN) based methods for object detection\, and then focus on t
he theoretical principles behind the regression-based YOLO approach. The h
ands-on sessions will be dedicated to actual parametrization\, training\,
and usage of such networks on classical datasets (PASCAL VOC\, COCO\, ...)
. Finally\, we will discuss how this method can be modified to predict add
itional parameters for each object\, still in the form of a single standal
one network\, and illustrate this capability for galaxy detection and char
acterization in radio-astronomical images.\n\n\n Generative models for ima
ges\, Bruno Galerne (Université d'Orléans)\n\n\nThe goal of this short c
ourse is to introduce the main deep generative models that have been devel
oped this last decade. These models are practical solutions for the unsupe
rvised learning problem of parametric modeling of any data distribution. T
he advances in deep learning representation have led to generative models
enable to generate synthetic realistic data.\n\nThe course will mainly foc
us on variational auto-encoders (VAE) and Generative Adversarial Networks
(GAN). The mathematical modeling will be presented and the basic propertie
s of fundamental tools for comparing distributions\, such as Kullback-Leib
ler divergences and optimal transport metrics\, will be recalled. Numerica
l examples will focus on image modeling although the range of applications
of these generic models is broader.\n\n\n Deep Learning and dynamical sys
tems: applications in neuroimaging\, François Rousseau (IMT Atlantique)\n
Introduction to deep learning on graphs\, Samuel Vaiter (CNRS\, Universit
é Côté d'Azur)\n\n\nThis course is an introduction to geometric deep le
arning with a focus on graphs and the mathematical aspects associated. Aft
er showing the main principles behind spectral and message passing graph n
eural networks\, the students will be confronted with the implementation o
f simple models with the help of PyTorch Geometric. The last part of the c
ourse will be dedicated to recents advances in the mathematical analysis o
f the large relatively sparse random regime.\n(C’est temporaire\, je ne
sais pas encore exactement ce que je mettrais dedans)\n\nTalks\n\n\n Rodri
go Ibata (Observatoire Astronomique de Strasbourg)\n Nicolas Padoy (IHU\,
Université de Strasbourg)\n\n\n\n \n\nhttps://indico.math.cnrs.fr/event/
8012/
LOCATION:IRMA Salle de conférence
URL:https://indico.math.cnrs.fr/event/8012/
END:VEVENT
END:VCALENDAR