Catálogo de publicaciones - libros

Multiple sequence alignment is a crucial technique for many fields of computational biology and remains a difficult task. Combining several different alignment techniques often leads to the best results in practice. Within this paper we present MAUSA (Multiple Alignment Using Simulated Annealing) and show that the conceptually simple approach of simulated annealing, when combined with a recent development in solving the aligning alignments problem, produces results which are competitive and in some cases superior to established methods for sequence alignment. We show that the application of simulated annealing to effective guide tree selection improves the quality of the alignments produced. In addition, we apply a method for the automatic assessment of alignment quality and show that in scenarios where MAUSA is selected as producing the best alignment from a suite of approaches (approximately 10% of test cases), it produces an average 5% (p = 0.005, Wilcoxon sign-rank test) improvement in quality.

The first source of motivation for our work is the modelling of multi-agent systems. In particular, we follow the approach of [1] that combines two perspectives: (a) a cognitive account of agents that specifies motivational attitudes, using the BDI architecture [2], and (b) modelling of agent societies by means of normative concepts [3].

This paper describes a translation of the time-reversal problem in to propositional satisfiability. Two useful features of this translation are: that the encoding is linear (in both variables and clauses) with respect to the number of cells in the original problem; and, it can be used to generate problem instances that are known to be satisfiable. The problem is shown to have statistically defined hard regions where instances are on average more difficult than in other regions.

Named entity disambiguation has been one of the main challenges to research in Information Extraction and development of Semantic Web. Therefore, it has attracted much research effort, with various methods introduced for different domains, scopes, and purposes. In this paper, we propose a new approach that is not limited to some entity classes and does not require well-structured texts. The novelty is that it exploits relations between co-occurring entities in a text as defined in a knowledge base for disambiguation. Combined with class weighting and coreference resolution, our knowledge-based method outperforms KIM system in this problem. Implemented algorithms and conducted experiments for the method are presented and discussed.

This paper proposes a local-search-based real-time scheduling method for non-crossing stacking cranes in an automated container terminal. Considering the dynamic property of the yard crane operation and real-time constraints, the method builds a new crane schedule for a fixed-length look-ahead horizon whenever a new crane job is requested. One difficulty in crane scheduling is that sometimes additional crane operations need to be done to complete a requested job, especially when other containers are stacked on top of the requested container. We use a redundant and variable-length representation of a candidate solution for search to accommodate those additional operations. Simulation experiment shows that the local-search-based method outperforms heuristic-based method in real-time situations.

The traditional concept of a genetic algorithm (GA) is that of selection, crossover and mutation. However, data from the literature has suggested that the niche for the beneficial effect of crossover upon GA performance may be smaller than has been traditionally held. We explored the class of problems for which crossover is detrimental by performing a statistical analysis of two test problem suites, one comprising linear-separable non-rotated functions and the other comprising the same functions rotated by 45 degrees rendering them not-linear-separable.

We find that for the rotated functions the crossover operator is detrimental to the performance of the GA. We conjecture that what makes a problem difficult for the GA is complex and involves factors such as the degree of optimization at local minima due to crossover, the bias associated with the mutation operator and the Hamming Distances present in the individual problems due to the encoding.

Finally, we test our GA on a practical landscape minimization problem to see if the results obtained match those from the difficult rotated functions. We find that they match and that the features which make certain of the test functions difficult are also present in the real world problem.

Research on automatic wood defect inspection has hardly studied sapstain. We propose to use machine learning techniques such as feature selection, visualization and classification, to build an automatic sapstain detection system.

In this paper, we discuss the structure based classification of documents based on their logical document structure, i.e., their DOM trees. We describe a method using predefined structural features and also four tree kernels suitable for such structures. We evaluate the methods experimentally on a corpus containing the DOM trees of newspaper articles, and on the well-known SUSANNE corpus. We will demonstrate that, for the two corpora, many text types can be learned based on structural features only.

This paper describes a neural-based technique for detecting upright persons in low-resolution beach imagery in order to predict trends of tourist activities at beach sites. The proposed system uses a structural feature extraction technique to represent objects of interest for training a selection of neural classifiers. A number of neural-based classifiers are compared in this study and a direction-based feature extraction technique is investigated in conjunction with a rotationally invariant procedure for the purpose of beach object classification. Encouraging results are presented for person detection using video imagery collected from a beach site on the coast of Australia.

The multilevel thresholding method with maximum entropy is one of the most important image segmentation methods in image processing. However, its time-consuming computation is often an obstacle in real time application systems. Particle swarm optimization (PSO) algorithm is a class of heuristic global optimization algorithms which appeared recently. In this paper, the maximum entropy is obtained through an adaptive particle swarm optimization (APSO) algorithm. The APSO algorithm is shown to obtain the maximum entropy of multilevel thresholding effectively on experiments of image segmentation.