Multi-target prediction (MTP) is concerned with the simultaneous prediction of multiple target variables of diverse type, such as binary, nominal, ordinal, real-valued or even mixed. Often, these multiple target variables are related either explicitly, for example they could represent a ranking, be nodes of a graph or have a spatial, temporal or spatiotemporal relationship, or implicitly, for example via hidden mutual exclusion or parent-child relationships.

While some progress in developing efficient and effective MTP methods has been achieved, we are still far from being able to successfully apply them to big data, an area which is expected to bring new and smart growth opportunities for Europe . Big data are often described via a number of "Vs", the most important ones for research being Volume, Velocity, Variety and Veracity. Complementary to existing big data efforts that address the "Vs" of input variables, the BigTargets workshop will focus on the "Vs" of the target (output) variables. Topics of interest include (but are not limited to):

• Efficient inference and large-scale learning in multi-target prediction with very large output spaces
• Data streams and real-time processing of multi-target variables
• Combining heterogeneous types of multi-target variables
• Combining heterogeneous sources of multi-target variables
• Uncertainty in multi-target variables (e.g. crowdsourcing, multiple human annotators, etc.)
• Evaluation of multi-target prediction systems
• Data sampling in multi-target prediction
• Theoretical results on multi-target prediction
• Incorporation of domain knowledge in multi-target prediction methods
• Application of scalable methods to real-world big multi-target prediction problems

As multi-target intends to unify several areas of machine learning, we welcome contributions from different domains:
• Multi-label classification
• Multivariate regression / Multi-output regression
• Structured output prediction
• Multi-task learning and transfer learning
• Constructive machine learning
• Pairwise learning / dyadic prediction
• Label ranking
• Matrix factorization and collaborative filtering methods
• Recommender systems
• Sequence learning, time series prediction and data stream mining
• Collective classification and inference

The program committee will make a selection for oral presentations among full papers and abstracts to ensure a program that has academic quality but is also interesting and inspiring for the attendees. Full papers can take up to 8 pages and they need to report original work that has not been published yet. Extended abstracts have a maximum of 2 pages and can also concern a discussion of a given topic or past published work, if the precise references to the original publication are mentioned. We strongly encourage people that want to give a talk to submit an abstract. Full papers are rather meant for researchers that need a publication for travel funding or for people that want to obtain more detailed feedback about their work. This way we aim to provide a broad overview, and make it attractive and easy to attend by both senior as well as junior researchers, from both academia and industry. Full papers and extended abstracts should be formatted according to the official ECML/PKDD style files. Accepted papers will be made available online at the workshop website, but the workshop proceedings can be considered non-archival. Submissions need not be anonymous. All papers should be submitted as pdf via email to multitargetprediction@gmail.com.