Learning is like a puzzle

The research activities of the lab are organised in the following streams.
Deep learning: the laboratory studies new learning paradigms that are able to define hierarchical architecture during the learning process.
Reinforcement learning: the laboratory deals with the definition of new reinforced learning schemes to facilitate learning based on experience and exploration in critical application contexts.
Active learning: the laboratory develops algorithms that are able to use the large amount of unlabeled data by actively quering the user / teacher during the learning phase in order to limit the need for labeled data.
Transfer learning: the laboratory investigates the topic of transfer learning with particular reference to the transfer of knowledge from large problems (Big Data) to problems in which the availability of data is limited (Small Data).
Unsupervised learning: The laboratory develops algorithms in pattern recognition and computer vision research fields to achieve the performance of the current supervised algorithms through unsupervised approaches.














Image of smoke detection through laser