We present a novel framework for learning to interpret and generate language using only perceptual context as supervision. We demonstrate its capabilities by developing a system that learns to sportscast simulated robot soccer games in both English and Korean without any language-specific prior knowledge. Training employs only ambiguous supervision consisting of a stream of descriptive textual comments and a sequence of events extracted from the simulation trace. The system simultaneously establishes correspondences between individual comments and the events that they describe while building a translation model that supports both parsing and generation. We also present a novel algorithm for learning which events are worth describing. Human evaluations of the generated commentaries indicate they are of reasonable quality and in some cases even on par with those produced by humans for our limited domain.

Decentralized planning in uncertain environments is a complex task generally dealt with by using a decision-theoretic approach, mainly through the framework of Decentralized Partially Observable Markov Decision Processes (DEC-POMDPs). Although DEC-POMDPS are a general and powerful modeling tool, solving them is a task with an overwhelming complexity that can be doubly exponential. In this paper, we study an alternate formulation of DEC-POMDPs relying on a sequence-form representation of policies. From this formulation, we show how to derive Mixed Integer Linear Programming (MILP) problems that, once solved, give exact optimal solutions to the DEC-POMDPs. We show that these MILPs can be derived either by using some combinatorial characteristics of the optimal solutions of the DEC-POMDPs or by using concepts borrowed from game theory. Through an experimental validation on classical test problems from the DEC-POMDP literature, we compare our approach to existing algorithms. Results show that mathematical programming outperforms dynamic programming but is less efficient than forward search, except for some particular problems. The main contributions of this work are the use of mathematical programming for DEC-POMDPs and a better understanding of DEC-POMDPs and of their solutions. Besides, we argue that our alternate representation of DEC-POMDPs could be helpful for designing novel algorithms looking for approximate solutions to DEC-POMDPs.

LEXSYS, (Legume Expert System) was a project conceived at IITA (International Institute of Tropical Agriculture) Ibadan Nigeria. It was initiated by the COMBS (Collaborative Group on Maize-Based Systems Research in the 1990. It was meant for a general framework for characterizing on-farm testing for technology design for sustainable cereal-based cropping system. LEXSYS is not a true expert system as the name would imply, but simply a user-friendly information system. This work is an attempt to give a formal representation of the existing system and then present areas where intelligent agent can be applied.

Probabilistic matrix factorization (PMF) is a powerful method for modeling data associated with pairwise relationships, finding use in collaborative filtering, computational biology, and document analysis, among other areas. In many domains, there is additional information that can assist in prediction. For example, when modeling movie ratings, we might know when the rating occurred, where the user lives, or what actors appear in the movie. It is difficult, however, to incorporate this side information into the PMF model. We propose a framework for incorporating side information by coupling together multiple PMF problems via Gaussian process priors. We replace scalar latent features with functions that vary over the space of side information. The GP priors on these functions require them to vary smoothly and share information. We successfully use this new method to predict the scores of professional basketball games, where side information about the venue and date of the game are relevant for the outcome.

Current statistical models for structured prediction make simplifying assumptions about the underlying output graph structure, such as assuming a low-order Markov chain, because exact inference becomes intractable as the tree-width of the underlying graph increases. Approximate inference algorithms, on the other hand, force one to trade off representational power with computational efficiency. In this paper, we propose two new types of probabilistic graphical models, large margin Boltzmann machines (LMBMs) and large margin sigmoid belief networks (LMSBNs), for structured prediction. LMSBNs in particular allow a very fast inference algorithm for arbitrary graph structures that runs in polynomial time with a high probability. This probability is data-distribution dependent and is maximized in learning. The new approach overcomes the representation-efficiency trade-off in previous models and allows fast structured prediction with complicated graph structures. We present results from applying a fully connected model to multi-label scene classification and demonstrate that the proposed approach can yield significant performance gains over current state-of-the-art methods.

This work-in-progress provides a preliminary cognitive investigation of how the external visualization of the Ternary diagram (TD) might be used as an underlying model for exploring the representation of simple 3D cuboids according to the theory of Conceptual Spaces. Gärdenfors introduced geometrical entities, known as conceptual spaces, for modeling concepts. He considered multidimensional spaces equipped with a range of similarity measures (such as metrics) and guided by criteria and mechanisms as a geometrical model for concept formation and management. Our work is inspired by the conceptual spaces approach and takes ternary diagrams as its underlying conceptual model. The main motivation for our work is twofold. First, Ternary Diagrams are powerful conceptual representations that have a solid historical and mathematical foundation. Second, the notion of overlaying an Information- Entropy function on a ternary diagram can lead to new insights into applications of reasoning about shape and other cognitive processes.

As the pace of Linked data generation and usage increases, so does the interest in intelligent, usable, and scalable browsing tools. Faceted browsing has potential to provide a foundation for effective dataset navigation. In this paper, we will discuss some of the anticipated benefits along with some associated challenges in building the next-generation faceted browsing system for the Web of Linked Data. We also present our initial system design and implementation.

The CausalityWorkbench project provides an environment to test causal discovery algorithms. Via a web portal, we provide a number of resources, including a repository of datasets, models, and software packages, and a virtual laboratory allowing users to benchmark causal discovery algorithms by performing virtual experiments to study artificial causal systems. We regularly organize competitions. In this paper, we explore the opportunities offered by development applications.

With the advent of RDFa and the at least partial support by major search engines, semantically structured data is more and more appearing on the Web. To enable high value use cases, links between entity descriptions need to be established. The linked data model suggests that links should be state explicitly by those who expose entity descriptions, but unlike on the normal web, incentives for doing so are unclear so that the model ultimately seems to fail in practice. In this position paper, we make the conjecture that explicit links are not needed for realizing the semantic web. We propose discuss how Record Linkage techniques are in general very well suited for the task but argue the need for a tool would allow data publishers to have an active role in producing entity descriptions that can then be linked automatically.

Research effort, with its emphasis on evaluation and measurable progress, things began to change. Instead SHRDLU (WIN72) is perhaps the canonical example. of systems whose architecture and vocabulary were The rapid growth of efforts to found the next generation of based on linguistic theory (in this case acoustic phonetics), systems on general-purpose knowledge representation languages new approaches based on statistical modelling and Bayesian (I'm thinking of several varieties of semantic nets, probability emerged and quickly spread. "Every time I fire a from plain to partitioned, as well as KRL, KL-ONE and linguist my system's performance improves" (Fred Jellinek, their successors, ending (not yet, of course) with CYC (See head of speech recognition at IBM, c. 1980, latterly repudiated (BRA08) for all these) stumbled to a halt once their failure by Fred but widely attested). As advanced from resolution theorem provers through a number more and more problems are re-conceived as instances of of stages to the current proliferation of a range of Description the noisy channel model, the empiricist paradigm continually Logic'reasoners'; Whereas in the 1970s and 1980s there grew, so did the need to manage the impact of change and was real energy and optimism at the interface between computational conflict: enter'truth maintenance', subsequently renamed and theoretical linguistics, the overwhelming success'reason maintenance'. While still using some of But outflanking these'normal science' advances of AI, the terminology of linguistic theory, computational linguistics the paradigm shifters were coming up fast on the outside: practioners are increasingly detached from theory itself, over the last ten years machine learning has spread from which has suffered a, perhaps connected, loss of energy and small specialist niches such as speech recognition to become sense of progress.