Orelia and research
ORELIA's technology refers to many designations : sound detection, automatic recognition of sounds, audio analysis, audio transcription, audio event recognition, audio pattern recognition, classification of sounds to name just a few.
Actually, ORELIA's research interests are on the interplay between machine learning, digital signal processing and computer science. We have developed a method for audio pattern discovery from extremely large-scale data sets, considering uncertainty and respecting computational constraints.
More precisely, our research focuses on pattern discovery tasks where the amount of data is extremely large (environmental sounds) and the solution desired to be sparse. An important challenge in the field of machine learning is to deal with the increasing amount of data that is available for learning and to leverage the (also increasing) diversity of information sources, describing these data.
Beyond classical vectorial data formats, data in the format of graphs, trees, strings and beyond have become widely available for data mining i.e. we deal with hierarchical structure of sounds that are categorized by humans.
Moreover, for interpretability, stability and economical reasons, we provide decision rules that rely on a small subset of the information sources and/or a small subset of the features describing the data.
Generalization is our graal.