Call for Papers
Neural computation and machine learning have revolutionized automatic data processing, opening the ground towards problems which cannot be modeled with classical statistical techniques. Interestingly, besides techniques which have its roots in mathematical frameworks such as the support vector machine, quite a few methods have been inspired by cognitive models such as deep networks, or reservoir computing. Still, there exist quite a few challenges in this domain: Not only the amount of data explodes in virtually all application areas but also their complexity with regard to dimensionality, structural variety, and multimodality. At the same time, the tasks become more and more complex, moving from simple classification or prediction in pattern recognition to involved learning scenarios in dynamic environments with no explicit single objective. Humans are capable of handling complex situations and tasks by means of a combination of different paradigms, whereas existing neural systems mostly mirror only one or a few facets of the whole game.A non-exhaustive list of topics tackled in the workshop is given by the following keywords:
- nonlinear dimensionality reduction, blind source separation, and visualization
- models for very large or streaming data sets
- parallelization and hardware implementations
- models for non-euclidean data
- recursive models and dynamic systems
- adaptive data representation
- bio-inspired models
- challenges in machine learning
- challenges in applications
