Freitas, Andre (DERI/Insight, NUI Galway) | Handschuh, Siegfried (University of Passau) | Curry, Edward (DERI/Insight)

The crisp/brittle semantic model behind databases limits the scale in which data consumers can query, explore, integrate and process structured data. Approaches aiming to provide more comprehensive semantic models for databases, which are purely logic-based (e.g. as in Semantic Web databases) have major scalability limitations in the acquisition of structured semantic and commonsense data. This work describes a complementary semantic model for databases which has semantic approximation at its center. This model uses distributional semantic models (DSMs) to extend structured data semantics. DSMs support the automatic construction of semantic and commonsense models from large-scale unstructured text and provides a simple model to analyze similarities in the structured data. The combination of distributional and structured data semantics provides a simple and promising solution to address the challenges associated with the interaction and processing of structured data.

Freitas, André (Digital Enterprise Research Institute (DERI)) | Silva, João C. P. da (Federal University of Rio de Janeiro) | O’Riain, Seán (Digital Enterprise Research Institute (DERI)) | Curry, Edward (Digital Enterprise Research Institute (DERI))

This work introduces distributional relational networks (DRNs), a knowledge representation (KR) framework which focuses on allowing semantic approximations over large-scale and heterogeneous knowledge bases. The proposed model uses the distributional semantics information embedded in large text/data corpora to provide a comprehensive and principled solution for semantic approximation. DRNs can be applied to open domain knowledge bases and can be used as a KR model for commonsense reasoning. Experimental results show the suitability of DRNs as a semantically flexible KR framework.

In experimental tests of human behavior in unstructured bargaining games, typically many joint utility outcomes are found to occur, not just one. This suggests we predict the outcome of such a game as a probability distribution. This is in contrast to what is conventionally done (e.g, in the Nash bargaining solution), which is predict a single outcome. We show how to translate Nash's bargaining axioms to provide a distribution over outcomes rather than a single outcome. We then prove that a subset of those axioms forces the distribution over utility outcomes to be a power-law distribution. Unlike Nash's original result, our result holds even if the feasible set is finite. When the feasible set is convex and comprehensive, the mode of the power law distribution is the Harsanyi bargaining solution, and if we require symmetry it is the Nash bargaining solution. However, in general these modes of the joint utility distribution are not the experimentalist's Bayes-optimal predictions for the joint utility. Nor are the bargains corresponding to the modes of those joint utility distributions the modes of the distribution over bargains in general, since more than one bargain may result in the same joint utility. After introducing distributional bargaining solution concepts, we show how an external regulator can use them to optimally design an unstructured bargaining scenario. Throughout we demonstrate our analysis in computational experiments involving flight rerouting negotiations in the National Airspace System. We emphasize that while our results are formulated for unstructured bargaining, they can also be used to make predictions for noncooperative games where the modeler knows the utility functions of the players over possible outcomes of the game, but does not know the move spaces the players use to determine those outcomes.

Qu, Chao, Mannor, Shie, Xu, Huan

We devise a distributional variant of gradient temporal-difference (TD) learning. Distributional reinforcement learning has been demonstrated to outperform the regular one in the recent study \citep{bellemare2017distributional}. In our paper, we design two new algorithms called distributional GTD2 and distributional TDC using the Cram{\'e}r distance on the distributional version of the Bellman error objective function, which inherits advantages of both the nonlinear gradient TD algorithms and the distributional RL approach. We prove the asymptotic almost-sure convergence to a local optimal solution for general smooth function approximators, which includes neural networks that have been widely used in recent study to solve the real-life RL problems. In each step, the computational complexity is linear w.r.t.\ the number of the parameters of the function approximator, thus can be implemented efficiently for neural networks.