Anđela Davidović - Deep Learning group - Research

About

The aim of the group is to build up on deep learning skills among the members of the group.

We have very diverse levels of expertise on the topic, from complete beginners to experts on the field. Most of the members are the engineers of the Hub of Bioinformatics and Biostatistics, but also from the Image Analysis Hub, and other departments of the institute including Hearing Institute.

We plan to organize monthly meetings where we will discuss different DL projects, most recent papers, organize hands-on sessions and tutorials.

If you are interested to join us or to stay informed, please send an e-mail to andela.davidovic@pasteur.fr or joseph.josephides@pasteur.fr.

DL Meeting #11 – May ’24

Who and what? – Bernd Jagla, Research Engineer at scBiomarkers UTechS & Bioinformatics and Biostatistics Hub of Institut Pasteur will talk about an ongoing journey of collaboration with Claude-3-7-sonnet, an Anthropic’s advanced large language model.
When and where? – Tuesday 13/05/2025 14:00, 1st floor C. François Jacob 01A (CFJ 28-01-01A) & Teams.
Organised by J. Josephides

DL Meeting #10 – April ’25

Who and what? – Ernest Mordret, post-doc in G5 Molecular diversity of microbes, that will talk to us about Large Language Models to reconstruct partially masked fragments of bacterial genomes.
When and where? – Tuesday 15/04, at 14h00 in LWOFF building, Salle Retrovirus RDC or Teams.
Organised by A. Davidovic

DL Meeting #9 – March ’25

Who and what?
- Jeremie Kalfon – PhD student from G5 Machine Learning for Integrative Genomics will talk about a AI based software scPRINT
- MD Christian Rose, Department of Emergency Medicine, Stanford University School of Medicine will talk about the future of AI in clinical medicine: promises, pitfalls, and a path Forward
When and where? – Wednesday 26/03, at 14h00 in LWOFF building, Salle Retrovirus RDC.
Organised by J. Josephides and A. Davidovic

DL Meeting #8 – October ’24

Who and what? – Joseph Josephides will show how machine learning helped us discover a new layer of genomic heterogeneity in cancer and explore the role DNA replication plays in tumour evolution.
When and where? – Wednesday 23/10, at 11h00 in CFJ building, Salle CFJ 28-01-01A – C. François Jacob 1er
Organised by J. Josephides and A. Davidovic

DL Meeting #7 – September ’24

Who and what? – Anna Zhukova from G5 Evolutionary dynamics of infectious diseases. She will tell us about the use of deep learning for parameter estimation and model selection from phylogenetic trees.
When and where? – Wednesday 18/09, at 15h30 in LWOFF building, Salle Retrovirus RDC.
Organised by A. Davidovic

DL Meeting #6 + Kick off discussion for 2024/2025 – July ’24

Who and what?
- Emma Soufir, master student of Thomas Bigot and Julia Kende from the bioinfo&bistat hub, will present her work on classification of viral reads using CNN and LSTM models.
- Kick-off discussion to plan the future of the group.
When and where? – Friday 12/07, at 15h00 in Yersin building, 24-RDC-02
Organised by A. Davidovic

DL Meeting #5 – June ’24

Who and what? – Guillaume BOUVIER, CR in the Structural Bioinformatics unit. He will tell us about InDeep, a tool for predicting functional binding sites within proteins. You will be able to learn about geometric deep learning and how to incorporate prior physical knowledge on the system into neural architectures.
When and where? – Thursday 20/06, at 15h00 in LWOFF building, Salle Retrovirus RDC
Organised by A. Davidovic

DL Meeting #4, Tutorial session! – April ’24

Who and what? – Sophie Bagur from the Hearing Institute will present a tutorial with hands-on session on Introduction to Deep Learning for audition data.
When? – Friday 12/04, at 13h30 -17h30
Where? – Yersin building (24), Salle de Formation, RDC
Organised by A. Davidovic

The tutorial is split into three parts :

a brief overview of AI history and introduction to the main concepts
a hands-on part in Python and/or Matlab in which the participants train a CNN to discriminate music from speech spectrograms
a hands-on part where the participants look at the output of the pre-trained network “Yamnet” that identifies the sounds present in snippets of spectrograms

DL Meeting #3 – November ’23

Who and what? – Konstantin Petrov (and Youssef Ghorbal). We will hear about resources available on cluster (the difference between GPUs, the limitations we can have, etc). We’ll also hear about frequent pitfalls and general ML/DL workloads running.
When and where? – Thursday 23/11, at 14h00 in CFJ building, Salle CFJ 28-01-01A – C. François Jacob 1er or remotely on Teams
Organised by A. Davidovic

DL Meeting #2 – October ’23

Who and what? – Nicolas Maillet and his work on the project involving identification of protein sequences
When and where? – Tuesday 24/10/23, at 14h00 in Yersin building, 24-RDC-02.
Organised by A. Davidovic

DL Meeting #1 – September ’23

Who and what? – Bernd Jagla will share with us his knowledge on neural networks, that he worked on during his PhD thesis, and his view on the field of deeplearning over the last 20 years.
When and where? – Thursday 21/09/23, at 15h30 in CFJ building, floor 1, room 01A.
Organised by A. Davidovic

Kick-off meeting