FEAST 2014 - Workshop on Features and Structures

Data consisting of an underlying structure (sets, graphs, trees, sequences...) associated with descriptive attributes is important in many fields such as as natural language processing, medical imaging, computer vision, bioinformatics, chemoinformatics, neuroscience, and social network analysis. Open problems include choice of representation, combining discrete structure with continuous features, and dealing with low sample size situations.
The purpose of this workshop is to provide an open, low-threshold forum for discussing such and related problems. We welcome both original contributions, preliminary ideas and overviews of work that has already been published elsewhere. To encourage high-quality submissions, oral presentations and posters will be selected based on an extended abstract.
We welcome contributions including but not limited to:
multiple instance learning
structural kernels (graphs, trees, sequences, sets, etc)
distance-based methods
embedding and metricity
topological descriptors
structured output
combining structural and feature-based classifiers
large-scale graph analysis
structures in computational topology and bioinformatics
benchmark datasets
applications