We argue for a compositional semantics grounded in a strongly typed ontology that reflects our commonsense view of the world and the way we talk about it. Assuming such a structure we show that the semantics of various natural language phenomena may become nearly trivial.

Max-product belief propagation is a local, iterative algorithm to find the mode/MAP estimate of a probability distribution. While it has been successfully employed in a wide variety of applications, there are relatively few theoretical guarantees of convergence and correctness for general loopy graphs that may have many short cycles. Of these, even fewer provide exact ``necessary and sufficient'' characterizations. In this paper we investigate the problem of using max-product to find the maximum weight matching in an arbitrary graph with edge weights. This is done by first constructing a probability distribution whose mode corresponds to the optimal matching, and then running max-product. Weighted matching can also be posed as an integer program, for which there is an LP relaxation. This relaxation is not always tight. In this paper we show that \begin{enumerate} \item If the LP relaxation is tight, then max-product always converges, and that too to the correct answer. \item If the LP relaxation is loose, then max-product does not converge. \end{enumerate} This provides an exact, data-dependent characterization of max-product performance, and a precise connection to LP relaxation, which is a well-studied optimization technique. Also, since LP relaxation is known to be tight for bipartite graphs, our results generalize other recent results on using max-product to find weighted matchings in bipartite graphs.

The semiring-based constraint satisfaction problems (semiring CSPs), proposed by Bistarelli, Montanari and Rossi \cite{BMR97}, is a very general framework of soft constraints. In this paper we propose an abstraction scheme for soft constraints that uses semiring homomorphism. To find optimal solutions of the concrete problem, the idea is, first working in the abstract problem and finding its optimal solutions, then using them to solve the concrete problem. In particular, we show that a mapping preserves optimal solutions if and only if it is an order-reflecting semiring homomorphism. Moreover, for a semiring homomorphism $\alpha$ and a problem $P$ over $S$, if $t$ is optimal in $\alpha(P)$, then there is an optimal solution $\bar{t}$ of $P$ such that $\bar{t}$ has the same value as $t$ in $\alpha(P)$.

Gaussian mixture models (GMM) and support vector machines (SVM) are introduced to classify faults in a population of cylindrical shells. The proposed procedures are tested on a population of 20 cylindrical shells and their performance is compared to the procedure, which uses multi-layer perceptrons (MLP). The modal properties extracted from vibration data are used to train the GMM, SVM and MLP. It is observed that the GMM produces 98%, SVM produces 94% classification accuracy while the MLP produces 88% classification rates.

Evolutionary Learning proceeds by evolving a population of classifiers, from which it generally returns (with some notable exceptions) the single best-of-run classifier as final result. In the meanwhile, Ensemble Learning, one of the most efficient approaches in supervised Machine Learning for the last decade, proceeds by building a population of diverse classifiers. Ensemble Learning with Evolutionary Computation thus receives increasing attention. The Evolutionary Ensemble Learning (EEL) approach presented in this paper features two contributions. First, a new fitness function, inspired by co-evolution and enforcing the classifier diversity, is presented. Further, a new selection criterion based on the classification margin is proposed. This criterion is used to extract the classifier ensemble from the final population only (Off-line) or incrementally along evolution (On-line). Experiments on a set of benchmark problems show that Off-line outperforms single-hypothesis evolutionary learning and state-of-art Boosting and generates smaller classifier ensembles.

Noise, corruptions and variations in face images can seriously hurt the performance of face recognition systems. To make such systems robust, multiclass neuralnetwork classifiers capable of learning from noisy data have been suggested. However on large face data sets such systems cannot provide the robustness at a high level. In this paper we explore a pairwise neural-network system as an alternative approach to improving the robustness of face recognition. In our experiments this approach is shown to outperform the multiclass neural-network system in terms of the predictive accuracy on the face images corrupted by noise.

One of the major problems in computational biology is the inability of existing classification models to incorporate expanding and new domain knowledge. This problem of static classification models is addressed in this paper by the introduction of incremental learning for problems in bioinformatics. Many machine learning tools have been applied to this problem using static machine learning structures such as neural networks or support vector machines that are unable to accommodate new information into their existing models. We utilize the fuzzy ARTMAP as an alternate machine learning system that has the ability of incrementally learning new data as it becomes available. The fuzzy ARTMAP is found to be comparable to many of the widespread machine learning systems. The use of an evolutionary strategy in the selection and combination of individual classifiers into an ensemble system, coupled with the incremental learning ability of the fuzzy ARTMAP is proven to be suitable as a pattern classifier. The algorithm presented is tested using data from the G-Coupled Protein Receptors Database and shows good accuracy of 83%. The system presented is also generally applicable, and can be used in problems in genomics and proteomics.

This paper proposes an approach to training rough set models using Bayesian framework trained using Markov Chain Monte Carlo (MCMC) method. The prior probabilities are constructed from the prior knowledge that good rough set models have fewer rules. Markov Chain Monte Carlo sampling is conducted through sampling in the rough set granule space and Metropolis algorithm is used as an acceptance criteria. The proposed method is tested to estimate the risk of HIV given demographic data. The results obtained shows that the proposed approach is able to achieve an average accuracy of 58% with the accuracy varying up to 66%. In addition the Bayesian rough set give the probabilities of the estimated HIV status as well as the linguistic rules describing how the demographic parameters drive the risk of HIV.

Web page ranking and collaborative filtering require the optimization of sophisticated performance measures. Current Support Vector approaches are unable to optimize them directly and focus on pairwise comparisons instead. We present a new approach which allows direct optimization of the relevant loss functions. This is achieved via structured estimation in Hilbert spaces. It is most related to Max-Margin-Markov networks optimization of multivariate performance measures. Key to our approach is that during training the ranking problem can be viewed as a linear assignment problem, which can be solved by the Hungarian Marriage algorithm. At test time, a sort operation is sufficient, as our algorithm assigns a relevance score to every (document, query) pair. Experiments show that the our algorithm is fast and that it works very well.

This paper considers the problem of reasoning on massive amounts of (possibly distributed) data. Presently, existing proposals show some limitations: {\em (i)} the quantity of data that can be handled contemporarily is limited, due to the fact that reasoning is generally carried out in main-memory; {\em (ii)} the interaction with external (and independent) DBMSs is not trivial and, in several cases, not allowed at all; {\em (iii)} the efficiency of present implementations is still not sufficient for their utilization in complex reasoning tasks involving massive amounts of data. This paper provides a contribution in this setting; it presents a new system, called DLV$^{DB}$, which aims to solve these problems. Moreover, the paper reports the results of a thorough experimental analysis we have carried out for comparing our system with several state-of-the-art systems (both logic and databases) on some classical deductive problems; the other tested systems are: LDL++, XSB, Smodels and three top-level commercial DBMSs. DLV$^{DB}$ significantly outperforms even the commercial Database Systems on recursive queries. To appear in Theory and Practice of Logic Programming (TPLP)