Artificial Intelligence

The LINCS is strongly involved in research related to AI, from the most theoretical aspects (learning theory, sample complexity, information-theoretic bounds) to the development of open-source software (see the Python package scikit-network for graph analysis) and the application to real use cases (cyber-security, anomaly detection, automatic classification of technical documents, content recommandation, text style transfer). We are interested in any type of data, but more especially in knowledge graphs, databases, texts, logs and time-series. Our research aims at making AI efficient, robust and explainable.


Data Mining

  • Clustering
  • Embedding/Representation learning
  • Hierarchical clustering
  • Metric learning
  • Knowledge graphs
  • Time series analysis
  • Causality inference

Foundations of Machine Learning

  • Possibilities and limitations of machine learning
  • Information theory and machine learning 
  • Active learning
  • Transfer learning
  • Online learning
  • Decentralized learning
  • Robustness and security of learning algorithms

Applications of AI

  • Traffic and performance prediction
  • Network design
  • Anomaly detection
  • Alarm logs
  • Localization
  • Social networks
  • Content recommendation
  • Personalized medicine
  • NLP
  • Predictive maintenance
  • Smart grids

Some LINCS members active on these topics

Photo Fullname Institution Office Homepage
Thomas Bonald  Institut Mines-Telecom 51 ?
Anne Bouillard  Nokia Bell Labs 31 ?
Elie de Panafieu  Nokia Bell Labs 42 ?
François Durand Nokia Bell Labs42
Philippe Jacquet Nokia Bell Labs42?
Leonardo Linguaglossa Institut Mines-Telecom23
Laurent Massoulié Inria30?
Dimitrios Milioris Nokia Bell Labs36?
Alexandre Proutière Inria30
Mauro Sozio Institut Mines-Telecom?
Other colleagues have activities related with IA, but not as a main topic.