Bertrand Lebichot

I am a PhD student specialized in data mining and machine learning. The thesis is currently on its last part, and I will be fully available in April. Therefore, I am also searching for new, more concrete, challenges.

I hold a Master of Engineering. I am currently a research assistant and part-time lecturer. I am fluent in French and English. I also speak Dutch and German.

Research fields

  • Network-based classification

    The classical feature space is replaced by / added to an adjacency matrix. Vector-based concepts must be translated to the graph world and the learning process consists in assigning a label for every unlabeled node.

  • Network criticality

    In network analysis, one important question is: Which node seems to be the most critical, or vital? To what extent the deletion of each node hurts the connectivity within the network in a broad sense?

  • Fraud detection

    Global card fraud losses amounted to Billion US dollars each year. We design collective inference algorithms to spread fraudulent influence through a network by using a limited set of confirmed fraudulent transactions.

  • Markov decision processes

    Optimal policy minimizes the expected cost of source-destination states, Markovian problems (costs are associated to decisions / actions). We propose an optimally randomized policy allowing a trade-off in exploration / exploitation.

  • Transfer learning

    Learning from different domains (recognize apples might be easier if we know how to recognize pears) or tasks (learning two or more task at the same time to build better features representation).

  • Deep learning

    Deep learning is a major trend in Machine Learning since a few years. We use the well-known Keras suite to tackle problems as classification, fraud detection, artificial intelligence,...


As teaching assistant

LECGE1215 Informatique en économie et gestion LINK
LINGE1225 Algorithmique et programmation en économie et gestion LINK
LSINF1250 Mathématique pour l’informatique LINK
LLSMF2013 Analyse de données quantitatives LINK
LSINF2275 Data mining and decision making LINK

As professor

MLSMM2154 Machine Learning LINK



Bertrand Lebichot. Network analysis based on bag-of-paths: classification, node criticality and randomized policies. PhD Thesis


Bertrand Lebichot, Kevin Francoisse, Ilkka Kivimaki and Marco Saerens. Semi-Supervised Classification through the Bag-of-Paths Group Betweenness. In IEEE Transactions on Neural Networks and Learning Systems volume 25 (6 2014), pp. 1173–1186.


Illka Kivimaki, Bertrand Lebichot, Jari Saramaki, Marco Saerens. Two Betweenness Centrality Measures based on Randomized Shortest Paths. In Scientific Reports. volume 6, Article number: 19668 (2016).


Bertrand Lebichot, Fabian Braun, Olivier Caelens and Marco Saerens. A graph-based, semi-supervised, credit card fraud detection system. In Cherifi H., Gaito S., Quattrociocchi W., Sala A. (eds) Complex Networks & Their Applications V. COMPLEX NETWORKS 2016. Studies in Computational Intelligence, vol 693. Springer, Cham.


Fabian Braun, Olivier Caelen, Evgueni Smirnov, Steven Kelk, Bertrand Lebichot. Improving Card Fraud Detection through Suspicious Pattern Discovery. In Benferhat S., Tabia K., Ali M. (eds) Advances in Artificial Intelligence: From Theory to Practice. IEA/AIE 2017. Lecture Notes in Computer Science, vol 10351. Springer, Cham.


Bertrand Lebichot and Marco Saerens. A Bag-of-Paths Node Criticality Measure. In Neurocomputing. Volume 275 (January 2018), pp. 224–236.


Bertrand Lebichot and Marco Saerens. An experimental study of graph-based semi-supervised classification with additional node information. Currently in reviewing (second round) in Information Fusion.

Bertrand Lebichot, Guillaume Guex, Ilkka Kivimaki and Marco Saerens. Constrained Randomized Shortest Path Problems. To be submitted soon.