PANAMA’s research activity is organized around three major research axes. A common thread of these three research axes is the need to design low-dimensional structured data models, to adjust the models to real data through learning and discovery, and to develop efficient algorithms to exploit these models on various tasks ranging from low-level signal acquisition to high-level information extraction and navigation.
PANAMA’s research axes primarily target the signal level. From this perspective, source separation will play a key role as one of PANAMA’s major applicative focus, with an established know-how and increasing industrial transfers generated by these activities.
In complement, PANAMA will strengthen the links between sparse representations, structure analysis and machine learning. This is expected
to yield decisive break-through in bridging the gap between the signal level and the semantic level for manipulating and exploiting large-size data collections. This will be achieved in synergy with the resources provided by the ERC Starting Grant PLEASE for the first four years of PANAMA.
-
Axis 1: sparse models and representations
-
Efficient sparse models and dictionary design for large-scale data
-
Compressive Learning
-
-
Axis 2: robust acoustic scene analysis
-
Compressive acquisition and processing of acoustic scenes
-
Robust audio source separation
-
-
Axis 3: large-scale audio content processing and self-organization
-
Motif discovery in audio data
-
Structure modeling and inference in audio and musical contents
-
