GCAI 2016 / 2nd Global Conference on Artificial Intelligence
previous day
next day
all days

View: session overviewtalk overview

11:00-12:30 Session 8: Reasoning and Knowledge
Towards a Decision Procedure for Uniform Interpolants in EL-TBoxes - a Proof-Theoretic Approach
SPEAKER: unknown

ABSTRACT. We propose a proof-theoretic decision procedure for subsumer interpolants of general TBoxes formulated in the Description Logic EL. A subsumer interpolant of a TBox for a signature and a concept is a TBox that only uses symbols from the signature and that entails exactly the subsumers over the signature of the concept as the original TBox. We show how the problem of deciding the existence of uniform interpolants of EL-TBoxes can be divided into three subproblems based on a characterisation of the logical difference between EL-TBoxes. The decision procedure for subsumer interpolants that we introduce in this paper solves one of these subproblems. We also evaluate our procedure by applying a prototype implementation on two biomedical ontologies.

Empirical Investigation of Learning-Based Imputation Policies
SPEAKER: unknown

ABSTRACT. Certain approaches for missing-data imputation propose the use of learning techniques to identify regularities and relations between attributes, which are subsequently used to impute some of the missing data. Prior theoretical results suggest that the soundness and completeness of such learning-based techniques can be improved by applying rules anew on the imputed data, as long as one is careful in choosing which rules to apply at each stage. This work presents an empirical investigation of three natural learning-based imputation policies: training rules once and applying them repeatedly; training new rules at each iteration; continuing the training of previous rules at each iteration. We examine how the three policies fare across different settings. In line with the predictions of the theory, we find that an iterative learn-predict approach is preferable.

When Should Learning Agents Switch to Explicit Knowledge?

ABSTRACT. According to psychological models, learned knowledge can be distinguished into implicit and explicit knowledge. The former can be exploited, but cannot be verbalized easily (e.g., to explain it to another person). The latter is available in an explicit form, it often comprises generalized, rule-based knowledge which can be verbalized and explained to others. During a learning process, the learned knowledge starts in an implicit form and gets explicit as the learning process progresses, and humans profit from exploiting such generalized, rule-based knowledge when learning. This paper investigates how learning agents can profit from explicit knowledge which is extracted during a learning process from a learned implicit representation. It is clearly shown that an agent can already profit from explicit knowledge in early phases of a learning process.

14:00-15:30 Session 9: Agents and Robots
A Short-Term Memory for Deliberative Agents in Everyday Environments
SPEAKER: unknown

ABSTRACT. Humans have the impressive capability to efficiently find near-optimal solutions to complex, multi-step problems. AI planning can model such problems well, but is inefficient for realistic problems. We propose to use AI planning in combination with a short-term memory, inspired by models of human short-term memory, to structure real-world problem domains and make the planning process more efficient, while still producing satisficing solutions. We evaluate the method in the domain of a household robot.

Heuristic Decision-Making for Human-aware Navigation in Domestic Environments

ABSTRACT. Robot navigation in domestic environments is still a challenge. This paper introduces a cognitively inspired decision-making method and an instantiation of it for (local) robot navigation in spatially constrained environments. We compare the method to two existing local planners with respect to efficiency, safety and legibility.

Cooperation in Adaptive Multi-Agent Systems through System of Systems modeling.
SPEAKER: unknown

ABSTRACT. This paper addresses the modeling and design of Systems of Systems (SoS) as well as inter multi-agent systems cooperation. It presents and illustrates a new generic model to describe formally SoS. Then, this model is used to propose a study of inter-AMAS (Adaptive Multi-Agent System) cooperation. Each AMAS, reified as a component-system of a SoS, uses a cooperative decision process in order to interact with other AMAS and collectively give rise to a relevant overall function at the SoS level. The proposed model as well as the inter-AMAS study are instantiated to a simulated problem of robots carrying colored boxes and compared to another heuristic decision.

16:00-18:00 Session 10: Applications II
Invariant Projections in Games

ABSTRACT. Identification of implicit structures in dynamic systems is a fundamental problem in Artificial Intelligence. In this paper, we focus on General Game Playing where games are modeled as finite state machines. We define a new property of game states called invariant projections which strongly corresponds to humans' intuition of game boards and may be applied in General Game Playing to support powerful heuristics, and to automate the design of game visualizations. We prove that the computation of invariant projections is Pi_{2}^{P}-complete in the size of the game description. We also show that invariant projections form a lattice, and the lattice ordering may be used to reduce the time to compute invariant projections potentially by a factor that is exponential in the schema size of game states. To enable competitive general game players to efficiently identify boards, we propose a sound (but incomplete) heuristic for computing invariant projections and evaluate its performance.

Heliostat Field Layout Optimization with Evolutionary Algorithms
SPEAKER: unknown

ABSTRACT. The exploitation of solar power for energy supply is of increasing importance. While technical development mainly takes place in the engineering disciplines, computer science offers adequate techniques for optimization. This work addresses the problem of finding an optimal heliostat field arrangement for a solar tower power plant. We propose a solution to this global, non-convex optimization problem by using an evolutionary algorithm. We show that the convergence rate of a conventional evolutionary algorithm is too slow, such that modifications of the recombination and mutation need to be tailored to the problem. This is achieved with a new genotype representation of the individuals.