Hello, I'm Thomas Robert

I’m a PhD student in Deep Learning working at the LIP6 lab (University Pierre and Marie Curie, Paris).

I started my PhD on November 2016, and I am supervised by Matthieu Cord and Nicolas Thome.


Our team

I am working at the Pierre and Marie Curie University in Paris, at the LIP6 lab, in the MLIA team managed by Patrick Gallinari. Our team, supervised by Matthieu Cord, is focusing on Computer Vision research and is of course taking the deep learning turn. Here are some example of publications by our team:

  • WILDCAT: Weakly Supervised Learning of Deep ConvNets for Image Classification, Pointwise Localization and Segmentation. T Durand, T Mordan, N Thome, M Cord. CVPR 2017.
  • LR-CNN for Fine-grained Classification with Varying Resolution., M Chevalier, N Thome, M Cord, J Fournier, G Henaff, E Dusch. ICIP 2015.
  • Recipe Recognition with Large Multimodal Food Dataset. X Wang, D Kumar, N Thome, M Cord, F Precioso ICMEW 2015.
  • MANTRA: Minimum Maximum Latent Structural SVM for Image Classification and Ranking. T Durand, N Thome, M Cord. ICCV 2015.
  • Sequentially Generated Instance-Dependent Image Representations for Classification. G Dulac-Arnold, L Denoyer, N Thome, M Cord, P Gallinari. ICLR 2014.
  • Fantope Regularization in Metric Learning. MT Law, N Thome, M Cord. CVPR 2014.
  • Learning deep hierarchical visual feature coding. H Goh, N Thome, M Cord, JH Lim. TNNLS 2014.
  • Top-Down Regularization of Deep Belief Networks. H Goh, N Thome, M Cord, JH Lim. NIPS 2013.
  • Pooling in image representation: The visual codeword point of view. S Avila, N Thome, M Cord, E Valle, ADA Araújo. CVIU 2013.
  • Quadruplet-wise Image Similarity Learning. M Law, N Thome, M Cord. ICCV 2013.
  • Unsupervised and Supervised Visual Codes with Restricted Boltzmann Machines. H Goh, N Thome, M Cord, JH Lim. ECCV 2012.

Research interests

My research interests are broad and goes from supervised deep learning for image classification to semantic representation learning for text and images and deep recurrent neural networks.

Currently, I am working on developing new architecture of convolutional neural networks for image classification using semi-supervised learning, and for this matter I have a particular interest on generative models of images (VAEs, GANs, etc.).

My skills & Hobbies

Data Science

  • Theoretical knowledge
  • Tensorflow / Keras
  • Torch
  • Scikit-learn
  • Matlab
  • Hadoop / Spark


  • Theoretical knowledge
  • Python
  • Java
  • Scala
  • C / C++
  • Lua
  • Bash
  • IDEs
  • SVN / Git


  • Theoretical knowledge
  • PHP
  • MySQL
  • JavaScript
  • HTML5
  • CSS3



  • Windows
  • OS X / Linux
  • Network
  • System Administration


  • LaTeX
  • Microsoft Office
  • OpenOffice



  • French
  • English
  • Spanish
  • German


  • Team work & Team Leadership
  • IT Project Management
  • Quality Management


News reading

  • Data Science & Big Data advances
  • IT news
  • General news
  • Scientists’ blogs

Books and series

  • Science-Fiction & Fantastic books
  • TV shows in original language
Basics / Beginner / Intermediate / Advanced

My experiences

Here is a timeline of my experiences.
Feel free to do explore it thanks to the controls on the left.

Click on a circle to toggle an item. Click on a label to focus on it only.

Type of experience:
  • Education
  • Professional
  • Associative
  • Personal project
  • Machine Learning
  • Research
  • Web
  • Big Data
  • Statistics
  • Business Intelligence
  • BPM
  • None
Show all
Machine Learning Research

PhD student in Deep Learning

LIP6 lab, Pierre and Marie Curie University – Paris, France

2017 Current

Supervised by Matthieu Cord and Nicolas Thome

Work on the DeepVision project in partnership with the LIRIS (INSA Lyon, France), the Simon Fraser University (Canada) and the University of Guelph (Canada).

Development of semi-supervised learning methods for deep CNNs (ongoing work)

Teaching assistant (64h / year) on statistical methods (PCA), neural networks for image classification, Java programming and more to come.

Machine Learning Research

Research engineer in Deep Learning

LIP6 lab, Pierre and Marie Curie University – Paris, France


Supervised by Matthieu Cord and Nicolas Thome

Work on various Machine Learning research problems using weakly supervised learning model and deep convolutional neural networks.

Research Web



DjLu is a simple and open source tool to organize for bibliographical library. It has been developed with Micael Carvalho and Rémi Cadène.

  • Tools used: HTML / CSS / Javascript / PHP / Git


Machine Learning Big Data Statistics Research

Data Science research internship

Conviva – San Francisco Bay, California, USA

2015 6 months

Work on various research problems about video streaming quality optimization using data science

  • Prediction of the optimal video quality for a given session using Machine Learning and A/B testing
  • Formulation and resolution of an optimization problem to determine optimal decisions Conviva's products should take
  • Statistical study of bandwidth evolution in the context of Conviva's products
  • Technologies: Python, Hive, Hadoop MapReduce, Spark
Machine Learning Statistics

Double degree in Data Science

National Institute of Applied Sciences & University of Rouen, France

2015 5 years

Master’s degree (“diplôme d’ingénieur”) in Data Science
Master’s degree in Multimedia Information Processing System

  • Preparatory cycle (2 years)
  • Engineering cycle in ASI department (3 years)
  • Master's complementary classes (1 year in parallel of INSA)
Machine Learning Big Data Statistics Research

Data-scientist intern

Creative Data – Rouen, France

Summer 2014 14 weeks

  • Self-training on Big Data concepts, architectures and technologies (Hadoop ecosystem).
  • Deployments of Big Data clusters on Google Cloud Platform.
  • Statistical work on the data of an e-commerce platform provider to give them some business intelligence insight about the shops they are hosting.
  • Benchmark of various Machine Learning algorithms to predict the next item of a sequence for a provider of portable self-scanning solution for retailers (use of Matlab & Python’s scikit-learn). Reading of Machine Learning literature.
Business Intelligence Web

Development of a web geo-analysis tool for SOLAP data - Academic project

For the Luxembourg Institute of Science and Technology (LIST), Luxembourg

2014 26 weeks 27h/week

  • Development of an open-source user-friendly web solution to analyze geographical SOLAP data over its various dimensions. Integration inside the GeoNode software.
  • Academic Research & Development project carried out for the LIST as client.
  • Project carried out with agile software development method (Scrum) with ISO 9001 quality standard.
  • Leader of a group of 7 students and responsible of relations with a subcontractor team of 8 students during half of the project. Developer during the other half.
  • Use of Git, JavaScript (d3.js, dc.js, crossfilter), Python (Django), Java, MDX, GeoMondrian.

Blog post Demo GitHub organization for the project




  • Contribution to the development of PGPic, a web tool for quality and project management.
  • Used by all PIC teams on a daily basis (PICs are professional projects such as TUDOR project presented above. There is 6 PIC projects each year.)
  • Developed skills: HTML / CSS3 / PHP / MySQL / Javascript

Analyst-developer intern

SOPRA Group – Insurance Agency – La Défense (Paris area), France

Summer 2013 9 weeks

  • In the domain of BPM (Business Process Modelling) mission of analyst-developer in a buisness context of employee savings management. Support of a senior business consultant.
  • Participation to work groups with the client (2 days a week) to determine their needs.
  • Modeling and optimization of core businesses processes in accordance with BMPN 2.0 standard.
  • Writing of detailled specifications of the activities.
  • Input of processes on the workflow software Bonita, and programming of forms and scripts to follow the process and validate the input data in accordance with the database. Use of Groovy and MySQL.
  • Creation of a web conversion tool beetween the output format of the process and the input format of the client's management software. Use of PHP, MySQL, jQuery, Bootstrap and FTP.

General Secretary

AJIR (Association Junior INSA Rouen) – Rouen, France

Feb. 2013 - Feb. 2014 1 year

AJIR is an association that puts in touch students and companies to carry out professional projects in engineering fields (IT, mechanics, energetics, chemistry, etc.).

  • Responsible of association and administrative management
  • Responsible of Board of Directors meetings
  • Humain Ressources Manager
  • Administrative paperwork
  • etc.

Website of AEI

Feb. 2013

  • Creation of a website for Association of INSA Students (AEI) throw AJIR with another student.
  • Experience of the realisation of a project for a real client.
  • Website created as a Wordpress template.
  • Developed skills: HTML / CSS3 / PHP / Javascript / Wordpress

See the website


Website of AJIR


  • Creation of the website for AJIR (Junior-Entreprise association) as a Wordpress template.
  • Creation of a plugin to manage membership workflow (quite heavy in Junior-Enterprises) in order to lighten HR work
  • Developed skills: HTML / CSS3 / PHP / Javascript / Wordpress

See the website

Chief Information Officer

AJIR (Association Junior INSA Rouen) – Rouen, France

Feb. 2012 - Feb. 2014 2 years

AJIR is an association that puts in touch students and companies to carry out professional projects in engineering fields (IT, mechanics, energetics, chemistry, etc.).

Support and improvement of the Information System of the Junior-Entreprise (sysadmin work & developpement of a new website).

Worker training period

Maintenance center of the French National Railway Company (SNCF) – Rouen, France

Summer 2011 1 month

Support of the brake team in its daily work. Discover of the working world.




  • Creation of a website simplifying the loan of books between students. Realized as a school project with a team.
  • Use of Facebook API for login and several API to get informations on the books from the ISBNs.
  • Relatively complex SQL queries.
  • Developed skills: HTML / CSS3 / PHP / MySQL / Javascript

See the website

Employee in the facility management department

Sogécap (Société Générale) – La Défense (Paris area), France

Summer 2010 1 month

Support of the facility management department (mail delivery in the office, orders, stock handing, etc.).

High-school scientific diploma (“baccalauréat scientifique”) with honors

Lycée d’Arsonval – Saint-Maur-des-Fossés (Paris area), France


Revision recaps


  • Writing (and publishing) of revision recaps of almost all my courses since last year of high-school until Master's degree.
  • Developed skills: Word 2010 / Inkscape

See the recaps


iGoogle plugin for BetaSeries


  • Plugin to manage a BetaSeries account (website to check which episodes of your series you have seen or not) for iGoogle.
  • Developed skills: HTML / CSS / PHP / XML / Javascript

Blog article


Who's who


  • Creation of a visual and interactive “who’s who” of the employees of Sogecap sorted alphabetically or by department.
  • Thought as simple as possible for HR: the pages are generated from pictures files named with people name and position in the compagny.
  • Cached HTML output for optimization.
  • Developed skills: HTML / CSS / PHP



  • Bilingual website listing themes and wallpapers for all video games consoles.
  • Important website allowing to read different contents (news, themes, wallpapers, surveys), each separated in categories.
  • Creation of an original design in accordance with Mediagen corporate identity.
  • Cache handling for performance.
  • Integration of a phpBB forum in the website.
  • Complete administration backend allowing:
    • Content production workflow (writing, correction, validation)
    • Smart files (themes and images) processing to reduce production time
    • Fine grain user access rights handling
    • Etc.
  • Developed skills: HTML / CSS / PHP / MySQL / Javascript