Shafer's theory of belief and the Bayesian theory of probability are two alternative and mutually inconsistent approaches toward modelling uncertainty in artificial intelligence. To help reduce the conflict between these two approaches, this paper reexamines expected utility theory-from which Bayesian probability theory is derived. Expected utility theory requires the decision maker to assign a utility to each decision conditioned on every possible event that might occur. But frequently the decision maker cannot foresee all the events that might occur, i.e., one of the possible events is the occurrence of an unforeseen event. So once we acknowledge the existence of unforeseen events, we need to develop some way of assigning utilities to decisions conditioned on unforeseen events. The commonsensical solution to this problem is to assign similar utilities to events which are similar. Implementing this commonsensical solution is equivalent to replacing Bayesian subjective probabilities over the space of foreseen and unforeseen events by random set theory probabilities over the space of foreseen events. This leads to an expected utility principle in which normalized variants of Shafer's commonalities play the role of subjective probabilities. Hence allowing for unforeseen events in decision analysis causes Bayesian probability theory to become much more similar to Shaferian theory.

In this paper some initial work towards a new approach to qualitative reasoning under uncertainty is presented. This method is not only applicable to qualitative probabilistic reasoning, as is the case with other methods, but also allows the qualitative propagation within networks of values based upon possibility theory and Dempster-Shafer evidence theory. The method is applied to two simple networks from which a large class of directed graphs may be constructed. The results of this analysis are used to compare the qualitative behaviour of the three major quantitative uncertainty handling formalisms, and to demonstrate that the qualitative integration of the formalisms is possible under certain assumptions.

Argumentation is the process of constructing arguments about propositions, and the assignment of statements of confidence to those propositions based on the nature and relative strength of their supporting arguments. The process is modelled as a labelled deductive system, in which propositions are doubly labelled with the grounds on which they are based and a representation of the confidence attached to the argument. Argument construction is captured by a generalized argument consequence relation based on the ^,--fragment of minimal logic. Arguments can be aggregated by a variety of numeric and symbolic flattening functions. This approach appears to shed light on the common logical structure of a variety of quantitative, qualitative and defeasible uncertainty calculi.

This paper presents and discusses several methods for reasoning from inconsistent knowledge bases. A so-called argumentative-consequence relation taking into account the existence of consistent arguments in favor of a conclusion and the absence of consistent arguments in favor of its contrary, is particularly investigated. Flat knowledge bases, i.e. without any priority between their elements, as well as prioritized ones where some elements are considered as more strongly entrenched than others are studied under different consequence relations. Lastly a paraconsistent-like treatment of prioritized knowledge bases is proposed, where both the level of entrenchment and the level of paraconsistency attached to a formula are propagated. The priority levels are handled in the framework of possibility theory.

Valuation networks have been proposed as graphical representations of valuation-based systems (VBSs). The VBS framework is able to capture many uncertainty calculi including probability theory, Dempster-Shafer's belief-function theory, Spohn's epistemic belief theory, and Zadeh's possibility theory. In this paper, we show how valuation networks encode conditional independence relations. For the probabilistic case, the class of probability models encoded by valuation networks includes undirected graph models, directed acyclic graph models, directed balloon graph models, and recursive causal graph models.

Relational models for diagnosis are based on a direct description of the association between disorders and manifestations. This type of model has been specially used and developed by Reggia and his co-workers in the late eighties as a basic starting point for approaching diagnosis problems. The paper proposes a new relational model which includes Reggia's model as a particular case and which allows for a more expressive representation of the observations and of the manifestations associated with disorders. The model distinguishes, i) between manifestations which are certainly absent and those which are not (yet) observed, and ii) between manifestations which cannot be caused by a given disorder and manifestations for which we do not know if they can or cannot be caused by this disorder. This new model, which can handle uncertainty in a non-probabilistic way, is based on possibility theory and so-called twofold fuzzy sets, previously introduced by the authors.

In the terminology of computer science, Concurrent Systems identify systems, either software, hardware or even biological systems, where sets of activities run in parallel with possible occasional interactions. A simple example of concurrent system is the Internet, which can be thought of as a set of computers, each one computing its independent activity, that often communicate to exchange some information. A further example is the railway system of a country, where many trains travel sharing tracks in an ordered way so that two trains can move at the same time along different tracks, whereas a single track (e.g, a platform in a train station) can only be used by a single train at a time. Furthermore, the large number of activities carried on by a single human cell form a biological concurrent system, that actually shares a number of similarities with the Internet. Compared to sequential systems, where a single action is executed at a time according to a sequential algorithm, concurrent systems raise new complex issues dealing with the ordering of action executions.

The best currently known interactive debugging systems rely upon some meta-information in terms of fault probabilities in order to improve their efficiency. However, misleading meta information might result in a dramatic decrease of the performance and its assessment is only possible a-posteriori. Consequently, as long as the actual fault is unknown, there is always some risk of suboptimal interactions. In this work we present a reinforcement learning strategy that continuously adapts its behavior depending on the performance achieved and minimizes the risk of using low-quality meta information. Therefore, this method is suitable for application scenarios where reliable prior fault estimates are difficult to obtain. Using diverse real-world knowledge bases, we show that the proposed interactive query strategy is scalable, features decent reaction time, and outperforms both entropy-based and no-risk strategies on average w.r.t. required amount of user interaction.

Bayesian network structure learning is the notoriously difficult problem of discovering a Bayesian network that optimally represents a given set of training data. In this paper we study the computational worst-case complexity of exact Bayesian network structure learning under graph theoretic restrictions on the (directed) super-structure. The super-structure is an undirected graph that contains as subgraphs the skeletons of solution networks. We introduce the directed super-structure as a natural generalization of its undirected counterpart. Our results apply to several variants of score-based Bayesian network structure learning where the score of a network decomposes into local scores of its nodes. Results: We show that exact Bayesian network structure learning can be carried out in non-uniform polynomial time if the super-structure has bounded treewidth, and in linear time if in addition the super-structure has bounded maximum degree. Furthermore, we show that if the directed super-structure is acyclic, then exact Bayesian network structure learning can be carried out in quadratic time. We complement these positive results with a number of hardness results. We show that both restrictions (treewidth and degree) are essential and cannot be dropped without loosing uniform polynomial time tractability (subject to a complexity-theoretic assumption). Similarly, exact Bayesian network structure learning remains NP-hard for "almost acyclic" directed super-structures. Furthermore, we show that the restrictions remain essential if we do not search for a globally optimal network but aim to improve a given network by means of at most k arc additions, arc deletions, or arc reversals (k-neighborhood local search).

This article addresses the issue of representing electroencephalographic (EEG) signals in an efficient way. While classical approaches use a fixed Gabor dictionary to analyze EEG signals, this article proposes a data-driven method to obtain an adapted dictionary. To reach an efficient dictionary learning, appropriate spatial and temporal modeling is required. Inter-channels links are taken into account in the spatial multivariate model, and shift-invariance is used for the temporal model. Multivariate learned kernels are informative (a few atoms code plentiful energy) and interpretable (the atoms can have a physiological meaning). Using real EEG data, the proposed method is shown to outperform the classical multichannel matching pursuit used with a Gabor dictionary, as measured by the representative power of the learned dictionary and its spatial flexibility. Moreover, dictionary learning can capture interpretable patterns: this ability is illustrated on real data, learning a P300 evoked potential. Keywords: Dictionary learning, orthogonal matching pursuit, multivariate, shift-invariance, EEG, evoked potentials, P300. 1. Introduction Scalp electroencephalography (EEG) measures electrical activity produced by post-synaptic potentials of large neuronal assemblies. Although this old medical imaging technique suffers from poor spatial resolution, EEG is still widely used in medical contexts (e.g. EEG devices are relatively cheap compared to other imaging techniques (e.g. MEG, fMRI, PET), and they offer both high temporal resolution (a short period of time between two acquisitions) and very low latency (a delay between the mental task and the recording on the electrodes).