Data Club

The Data Club is opened to PhD candidates, master degree students and post-docs working in the data field and who are interested in promoting interdisciplinarity and interacting with the private sectors. There are no more activities for the moment.

The Data Club brings together PhD candidates, master degree students and post-docs from a diversity of fields and labs : IGE, CESICE, TIMC, Gipsa-lab, etc.

You want to receive information about the activites and events organized by the Data Club ? Register to the mailing list
There are no more activities for the moment. 
You want to be involved in the Data Club ? Join the team !

Each month, the Data Club organized sessions of 2 hours, one Wednesday per month, from 4 to 6 pm, at the IMAG building.

Presentations made by young researchers will last 30 to 40 min, on Data topics that can be useful to many people. It will be followed by 20 min for discussion. Than, we will have 1 hour for refreshment and network.
We welcome potential speakers for the monthly talks. If you are interested, please contact Ferielle Podgorski

Upcoming sessions
Date Title Speakers
 February 28, 2018  Academic data science: conducting research at the interface of different disciplines  Carlos Gomez
 The proliferation of Data Science Institutes/Centers within major universities is tightly connected to the explosion of “data science” in industry and business during the past few years. What was a buzzword not long ago might have a transformative effect in modern science. One of the main goals of these institutes, such as our own at the UGA, is to foster frequent and long-lasting collaborations between different fields: sociology, astronomy, geophysics among others, on one hand, and computer science, machine learning, statistics on the other.
During the first half of my presentation, I will describe my work on image analysis and machine learning for astronomy, and the study of extra-solar planets via direct imaging. I will show how my doctoral thesis has been enabled by an interdisciplinary approach from day one. Then, I will discuss the many challenges that, from my point of view, we face as academic data scientists working at the interface of several disciplines. Finally, I will conclude by listing some of the exciting opportunities that data science opens for us, young researchers, and our future (perhaps academic) careers.
Slides of the presentation
 March 28, 2018  Delving deep in the ocean with deep learning  Redouane Iguensat
 Unless you were living under a rock, Deep Learning (DL) is definitely « the » hot topic in the interface between Computer Science and Mathematics. It is becoming the backbone of most Artificial Intelligence (AI) oriented applications and is leading to impressive results in many fields. Image/Voice recognition, Medical Imaging, Law, Entertainment, Game solving, etc… are some examples where DL is doing great currently. The first part of this talk will be dedicated to a general presentation to DL (beginner friendly).

The second part of the talk aims to give a glimpse to what is like to « transfer » the DL knowledge to a very special field: Spatial Oceanography. I would like to show you the challenges when using methods from a field in another, the importance of getting out of your scientific comfort zone, and in the end, share with you my experience that I hope can benefit to anyone who wants to apply DL to physical sciences.

Register to this session (Indicate your full name)
 May 3, 2018  Data protection challenges : placing technology in service of humanity  Katia Bouslimani & Ludovica Robustelli
 #DeleteFacebook, Edward Snowden, Wikileaks … Personal data protection currently represents a big challenge. The recent scandal involving Facebook and Cambridge Analytica proves once again that there are still some dangers affecting data protection. Our presentation will focus on two main issues: dangers connected the processing of personal data and legal protection of personal data, in particular legal instruments of protection available in the European Union.

We will then present our researches about personal data protection. How can we adapt the already existing legal instruments to the Internet? Is it still possible to protect ourselves from massive data mining and profiling?

Register to this session (Indicate your full name)
 May 30, 2018
 Graphical models and ecology : linking ecological theory to statistical models  Marc Ohlmann
Biogeography is a science that had its foundations in the scientific expeditions of Von Humboldt in Central America at the beginning of the XIXth century. It aims to describe and understand the patterns of species spatial distributions. Biotic interactions between organisms have long been ignored in biogeographic models partially because sampling biotic interactions is time consuming. However, they might be of prime importance in understanding and predicting biodiversity. Inferring species interaction network from spatial data (ie presence absence of species in multiple locations) is still an open questions. Several statistical tools around graphical models and especially bayesian networks have been explored to infer species interaction network. Nevertheless, analysing the structure of such objects using classic theory (i.e. theory built using continuous time dynamical models like Lotka-Volterra model) is confusing since they do not represent the same models. In other words, we aim to link ecological theory with statistical models. We propose to generalize an existing ecological theory (Trophic Theory of Island Biogeography) using several models of dynamic Bayesian networks. We will explore stationary distributions of theses objects. We will emphasize on the link between the structure of the network and the expected number of species observe using both simulated and ecological data.

Register to this session  (Indicate your full name)
 June 27, 2018

High-dimensional data: a different kind of big data

 Florian Privé
In Bioinformatics, many data are ultra high-dimensional matrices, having up to tens of millions of variables. Special algorithms have to be used to get information from those data.
Moreover, those matrices now have also a lot of rows (up to one million individuals), which makes them very big and more difficult to analyze.

Register to this session (Indicate your full name)

The Data Club will also organize events with companies.
You are from a company and you would like to meet students working on data? Please contact Ferielle Podgorski

Published on October 26, 2020