- Study of the multi-label learning problem and its related topics (label correlation, dimensionality reduction, feature selection, etc).
- Study of the emerging interactive learning problem (active learning, online learning, concept drift, etc).
- Definition of the interactive multi-label learning.
- Extensive study of the predictive and computation-time performances of multi-label learning algorithms with interactivity constraints (learning from few training examples in a very short time).
- Benchmarking with three common multi-label librairies: MeKa, MULAN and CLUS.
- Study of the multi-label dimensionality reduction approaches.
- Study of the use of Matrix Factorization approac...