Machine Learning for Big Data
In recent years, Big Data has become a new challenge that includes capture, storage, search, transfer, analysis. and visualization. Scientists regularly encounter limitations due to large data sets in many areas, including physics, genomics, biological and environmental research to cite a few. The aim of this special session is to present new methods and strategies to process data sets with sizes beyond the ability of commonly used software tools.
Organized by Cédric Richard
Optimization Techniques for Large-Scale Learning
Over the last decade, the need to analyze growing amounts of data have raised an increasing interest in the development of new optimization techniques. One of the key challenges is to introduce algorithms allowing us to solve machine learning problems involving massive and possibly distributed data sets. A first leverage to achieve scalability is to handle data as a flow, in an "online" fashion. A second one is to design methods that lead themselves to a parallel or distributed implementation over a network. This session considers recent advances related to these two aspects.
Organized by Pascal Bianchi
Bayesian Nonparametric Methods for Machine Learning
Bayesian nonparametrics has seen major advances in statistics and machine learning. Their theory and methods continue to grow, which inter alia, offers novel opportunities for modeling and processing of signals. The rising popularity of Bayesian nonparametric methods in signal processing is due to the flexibility they offer in processing signals and the wide range of areas where they can be applied. This special session presents some of the latest advances in Bayesian nonparametrics in signal processing.