Machine learning algorithms, despite their increasing success in handling object recognition tasks, still seldom perform without error. Often the process of understanding why the algorithm has failed is the task of the human who, using domain knowledge and contextual information, can discover systematic shortcomings in either the data or the algorithm. This paper presents an approach where the process of reasoning about errors emerging from a machine learning framework is automated using symbolic techniques. By utilizing spatial and geometrical reasoning between objects in a scene, the system is able to describe misclassified regions in relation to its context. The system is demonstrated in the remote sensing domain where objects and entities are detected in satellite images.
Among the local consistency techniques used for solving constraint networks, path-consistency (PC) has received a great deal of attention. However, enforcing PC is computationally expensive and sometimes unnecessary. Directional PC (DPC) is a weaker notion of PC that considers a given variable ordering and can thus be enforced more efficiently than PC. This paper shows that (the DPC enforcing algorithm of Dechter and Pearl) decides the constraint satisfaction problem (CSP) of a constraint language if it is complete and has the variable elimination property (VEP). However, we also show that no complete VEP constraint language can have a domain with more than two values. We then present a simple variant of the algorithm, called, and show that the CSP of a constraint language can be decided by if it is closed under a majority operation. In fact, is sufficient for guaranteeing backtrack-free search for such constraint networks. Examples of majority-closed constraint classes include the classes of connected row-convex constraints and tree-preserving constraints, which have found applications in various domains, such as scene labeling, temporal reasoning, geometric reasoning and logical filtering. Our experimental evaluations show that significantly outperforms the state- of-the-art algorithms for solving majority-closed constraints.
We introduce, study, and evaluate a novel algorithm in the context of qualitative constraint-based spatial and temporal reasoning that is based on the idea of variable elimination, a simple and general exact inference approach in probabilistic graphical models. Given a qualitative constraint network N, our algorithm utilizes a particular directional local consistency, which we denote by (sic)-consistency, in order to efficiently decide the satisfiability of N. Our discussion is restricted to distributive subclasses of relations, i.e., sets of relations closed under converse, intersection, and weak composition and for which weak composition distributes over non-empty intersections for all of their relations. We demonstrate that enforcing (sic)-consistency in a given qualitative constraint network defined over a distributive subclass of relations allows us to decide its satisfiability, and obtain similar useful results for the problems of minimal labelling and redundancy. Further, we present a generic method that allows extracting a scenario from a satisfiable network, i.e., an atomic satisfiable subnetwork of that network, in a very simple and effective manner. The experimentation that we have conducted with random and real-world qualitative constraint networks defined over a distributive subclass of relations of the Region Connection Calculus and the Interval Algebra, shows that our approach exhibits unparalleled performance against state-of-the-art approaches for checking the satisfiability of such constraint networks.
Partial singleton (weak) path consistency, or partial lozenge-consistency, for a qualitative constraint network, ensures that the process of instantiating any constraint of that network with any of its base relations b and enforcing partial (weak) path consistency, or partial lozenge-consistency, in the updated network, yields a partially lozenge-consistent subnetwork where the respective constraint is still defined by b. This local consistency is essential for helping to decide the satisfiability of challenging qualitative constraint networks and has been shown to play a crucial role in tackling more demanding problems associated with a given qualitative constraint network, such as the problem of minimal labeling. One of the main downsides to using partial lozenge-consistency, is that it is computationally expensive to enforce in a given qualitative constraint network, as, despite being a local consistency in principle, it retains a global scope of the network at hand. In this paper, we propose a lazy algorithm that restricts the singleton checks associated with partial lozenge-consistency to constraints that are likely to lead to the removal of a base relation upon their propagation. A key feature of this algorithm is that it collectively eliminates certain unfeasible base relations by exploiting singleton checks. Further, we show that the closure that is obtained by our algorithm is incomparable to the one that is entailed by partial lozenge-consistency and non-unique in general. We demonstrate the efficiency of our algorithm via an experimental evaluation with random Interval Algebra networks from the phase transition region of two separate models and, moreover, show that it can exhibit very similar pruning capability for such networks to the one of an algorithm for enforcing partial lozenge-consistency.
Addressing the lack of social, cognitive, and physical stimuli among elders is a key factor to contrast Mild Cognitive Impairment (MCI) that can arise during the third age. Against such background, agent-based technology has been applied to different application domains related to the assistance of elders. In this demo, we introduce an application of this kind: an activity center featuring social, cognitive, and physical activities targeted for elders. This activity center interacts with an autonomous agent, called Virtual Caregiver, residing in the cloud and generating interventions based on users’ data. We show how the user experience can be enriched with an adaptive configuration encouraging socialization and cognitive training.