The aim of the workshop is to facilitate communication between mathematicians of statistics and probability  and computer scientists on the topic of imprecise probabilities. This topic has witnessed a considerable development in the last 25 years, in connection with artificial intelligence statistics and decision theory, but it remains a confidential area in France, and completely absent from mathematical departments. The main point of imprecise probability is to augment the probabilistic approach to information processing with the capability of handling incomplete information, ambiguity, epistemic uncertainty in a non-biased way. One issue is to propagate partial ignorance through computation. It is especially of interesting areas where data is scarce and expert information is possibly all that is available.  It also addresses situations where the choice of a prior probability is problematic and when its choice would affect the validity of result of a computation. The general mathematical framework is that of convex sets of probabilities, and it includes as a special case random sets (in the form of belief functions), more dedicated to set-valued representations of imprecise data, as well as possibility theory, much developed at IRIT in the last 30 years. The latter has close connection with some statistical tools such as likelihood functions, confidence intervals and probabilistic inequalities.

The workshop brings together European specialists of imprecise probability methods in statistics, learning, stochastic processes and decision processes, along with more classical statisticians from IMT, Toulouse, as well as researchers from IRIT that are involved in imprecise probability research, with a special emphasis on machine learning and decision processes. It also includes application-oriented talks on the use of imprecise probability in signal processing and risk analysis.

Among expected benefits for participants of the workshop are
- A better understanding between IMT and IRIT on topics related to information representation and learning that may foster collaborations between scientists of each laboratory.
- A deeper understanding of imprecise probability theory and what it can be useful for, especially for the people that attended the course of Inés Couso in January.