Evaluation of pronoun resolution algorithm for Spanish dialogues

In this paper, we present an algorithm for anaphora resolution in Spanish dialogues and an evaluation of the algorithm for pronominal anaphora. The proposed algorithm uses both linguistic information and the structure of the dialogue to find the antecedent of the anaphors. The system has been evaluated on ten dialogues.

A novel hybrid simulation-optimization approach for the optimal design of multicomponent distillation columns

Póster presentado en Escape 22, European Symposium on Computer Aided Process Engineering, University College London, UK, 17-20 June 2012.

A hybrid simulation-optimization approach for the design of internally heat-integrated distillation columns

Paper submitted to AIChE 2012 Annual Meeting: Energy Efficiency by Process Intensification, Pittsburgh, PA, October 28-November 2, 2012.

Evaluating approximations generated by the GNG3D method for mesh simplification

In this paper we present different error measurements with the aim to evaluate the quality of the approximations generated by the GNG3D method for mesh simplification. The first phase of this method consists on the execution of the GNG3D algorithm, described in the paper. The primary goal of this phase is to obtain a simplified set of vertices representing the best approximation of the original 3D object. In the reconstruction phase we use the information provided by the optimization algorithm to reconstruct the faces thus obtaining the optimized mesh. The implementation of three error functions, named Eavg, Emax, Esur, permitts us to control the error of the simplified model, as it is shown in the examples studied.

Computational cost of GNG3D algorithm for mesh simplification

In this paper we present a study of the computational cost of the GNG3D algorithm for mesh optimization. This algorithm has been implemented taking as a basis a new method which is based on neural networks and consists on two differentiated phases: an optimization phase and a reconstruction phase. The optimization phase is developed applying an optimization algorithm based on the Growing Neural Gas model, which constitutes an unsupervised incremental clustering algorithm. The primary goal of this phase is to obtain a simplified set of vertices representing the best approximation of the original 3D object. In the reconstruction phase we use the information provided by the optimization algorithm to reconstruct the faces thus obtaining the optimized mesh. The computational cost of both phases is calculated, showing some examples.

Grouping genetic operators for the delineation of functional areas based on spatial interaction

The delineation of functional economic areas, or market areas, is a problem of high practical relevance, since the delineation of functional sets such as economic areas in the US, Travel-to-Work Areas in the United Kingdom, and their counterparts in other OECD countries are the basis of many statistical operations and policy making decisions at local level. This is a combinatorial optimisation problem defined as the partition of a given set of indivisible spatial units (covering a territory) into regions characterised by being (a) self-contained and (b) cohesive, in terms of spatial interaction data (flows, relationships). Usually, each region must reach a minimum size and self-containment level, and must be continuous. Although these optimisation problems have been typically solved through greedy methods, a recent strand of the literature in this field has been concerned with the use of evolutionary algorithms with ad hoc operators. Although these algorithms have proved to be successful in improving the results of some of the more widely applied official procedures, they are so time consuming that cannot be applied directly to solve real-world problems. In this paper we propose a new set of group-based mutation operators, featuring general operations over disjoint groups, tailored to ensure that all the constraints are respected during the operation to improve efficiency. A comparative analysis of our results with those from previous approaches shows that the proposed algorithm systematically improves them in terms of both quality and processing time, something of crucial relevance since it allows dealing with most large, real-world problems in reasonable time.

Growing Neural Gas approach for obtaining homogeneous maps by restricting the insertion of new nodes

The Growing Neural Gas model is used widely in artificial neural networks. However, its application is limited in some contexts by the proliferation of nodes in dense areas of the input space. In this study, we introduce some modifications to address this problem by imposing three restrictions on the insertion of new nodes. Each restriction aims to maintain the homogeneous values of selected criteria. One criterion is related to the square error of classification and an alternative approach is proposed for avoiding additional computational costs. Three parameters are added that allow the regulation of the restriction criteria. The resulting algorithm allows models to be obtained that suit specific needs by specifying meaningful parameters.

Logic hybrid simulation-optimization algorithm for distillation design

In this paper, we propose a novel algorithm for the rigorous design of distillation columns that integrates a process simulator in a generalized disjunctive programming formulation. The optimal distillation column, or column sequence, is obtained by selecting, for each column section, among a set of column sections with different number of theoretical trays. The selection of thermodynamic models, properties estimation etc., are all in the simulation environment. All the numerical issues related to the convergence of distillation columns (or column sections) are also maintained in the simulation environment. The model is formulated as a Generalized Disjunctive Programming (GDP) problem and solved using the logic based outer approximation algorithm without MINLP reformulation. Some examples involving from a single column to thermally coupled sequence or extractive distillation shows the performance of the new algorithm.

Logic-Based Outer-Approximation Algorithm for Solving Discrete-Continuous Dynamic Optimization Problems

We present an extension of the logic outer-approximation algorithm for dealing with disjunctive discrete-continuous optimal control problems whose dynamic behavior is modeled in terms of differential-algebraic equations. Although the proposed algorithm can be applied to a wide variety of discrete-continuous optimal control problems, we are mainly interested in problems where disjunctions are also present. Disjunctions are included to take into account only certain parts of the underlying model which become relevant under some processing conditions. By doing so the numerical robustness of the optimization algorithm improves since those parts of the model that are not active are discarded leading to a reduced size problem and avoiding potential model singularities. We test the proposed algorithm using three examples of different complex dynamic behavior. In all the case studies the number of iterations and the computational effort required to obtain the optimal solutions is modest and the solutions are relatively easy to find.

Osmium impregnated on magnetite as a heterogeneous catalyst for the syn-dihydroxylation of alkenes

A new catalyst derived from osmium has been prepared, fully characterized and tested in the dihydroxylation of alkenes. The catalyst was prepared by wet impregnation methodology of OsCl3·3H2O on a commercial micro-magnetite surface. The catalyst allowed the reaction with one of the lowest osmium loadings for a heterogeneous catalyst and was selective for the monodihydroxylation of 1,5-dienes. Moreover, the catalyst was easily removed from the reaction medium by the simple use of a magnet. The selectivity of catalyst is very high with conversions up to 99%. Preliminary kinetics studies showed a first-order reaction rate with respect to the catalyst.

3D model reconstruction using neural gas accelerated on GPU

In this work, we propose the use of the neural gas (NG), a neural network that uses an unsupervised Competitive Hebbian Learning (CHL) rule, to develop a reverse engineering process. This is a simple and accurate method to reconstruct objects from point clouds obtained from multiple overlapping views using low-cost sensors. In contrast to other methods that may need several stages that include downsampling, noise filtering and many other tasks, the NG automatically obtains the 3D model of the scanned objects. To demonstrate the validity of our proposal we tested our method with several models and performed a study of the neural network parameterization computing the quality of representation and also comparing results with other neural methods like growing neural gas and Kohonen maps or classical methods like Voxel Grid. We also reconstructed models acquired by low cost sensors that can be used in virtual and augmented reality environments for redesign or manipulation purposes. Since the NG algorithm has a strong computational cost we propose its acceleration. We have redesigned and implemented the NG learning algorithm to fit it onto Graphics Processing Units using CUDA. A speed-up of 180× faster is obtained compared to the sequential CPU version.

Global behavior of the Douglas–Rachford method for a nonconvex feasibility problem

In recent times the Douglas–Rachford algorithm has been observed empirically to solve a variety of nonconvex feasibility problems including those of a combinatorial nature. For many of these problems current theory is not sufficient to explain this observed success and is mainly concerned with questions of local convergence. In this paper we analyze global behavior of the method for finding a point in the intersection of a half-space and a potentially non-convex set which is assumed to satisfy a well-quasi-ordering property or a property weaker than compactness. In particular, the special case in which the second set is finite is covered by our framework and provides a prototypical setting for combinatorial optimization problems.

The Common Security and Defence Policy: National Perspectives. Egmont Paper No. 79, May 2015

FOREWORD. When one looks at the present state of the CSDP, one cannot help but look on with disenchantment at the energy that appears to have abandoned both institutions and Member States. Commentators increasingly take for granted that nothing much should be expected from this field of EU policy. The reasons for this state of mind are well known: the recent economic and financial strains, which have impacted all EU action since 2008, means that most of the Member States will struggle to keep their defence budgets at their present level in the future, and we may even see reductions. Furthermore, and to put it mildly, most of the recent CSDP operations have also experienced a lack of enthusiasm. Adding to this overall trend, the EU is far from presenting a common vision of what security and defence should really mean. Many of the Member States do not want to be involved in all of today’s international turmoils, and they rarely share the strategic culture which inspires those Member States who see themselves as having special responsibilities in dealing with these crises. In the end it may be that Member States diverge fundamentally on the simple question of whether it is relevant for the EU to engage in most of the ‘hot’ crises Europe faces; many prefer to see Europe as a soft power, mostly dedicated to intervening on less dramatic fronts and more inclined to mend than to fight. For whatever reason given, it remains that if there is a lack of common understanding on what CSDP should really be about, it should not come as a surprise if this policy is presently in stalemate. As an additional blow, the Ukrainian crisis, which dragged on for the whole of last year, could only add to the downward spiral the EU has been experiencing, with a new Russia aggressively confronting Europe in a manner not too distant from the Cold War days. This attitude has triggered the natural reaction among EU Member States to seek reassurances from NATO about their own national security. Coupled with the return of France a few years ago into the integrated military command, NATO’s renewed relevance has sent a strong message to Europe about the military organisation’s credibility with regard to collective defence. Surprisingly, this overall trend was gathering momentum at the same time as other more positive developments. The European Council of December 2013 dedicated its main session to CSDP: it underlined Europe’s role as a ‘security provider’ while adopting a very ambitious road map for Europe in all possible dimensions of the security sector. Hence the impression of a genuine boost to all EU institutions, which have been invited to join efforts and give CSDP a reinvigorated efficiency. In the same way, the increasing instability in Europe’s neighbourhood has also called for more EU operations: most recently in Iraq, Libya, Northern Nigeria or South Sudan. Pressure for further EU engagement has been one of the most constant features of the discussions taking place around these crises. Moreover, a growing number of EU partners in Asia, Latin America or Eastern Europe have shown a renewed eagerness to join CSDP missions in what sounds like a vote of confidence for EU capacities. What kind of conclusion should be drawn from this contradictory situation? Probably that the EU has much more potential than it can sometimes figure out itself, if only it would be ready to adapt to the new global realities. But, more than anything else, an enhanced CSDP needs from all Member States strong political will and a clear vision of what they want this policy to be. Without this indispensable ingredient CSDP may continue to run its course, as it does today. It may even grow in efficiency but it will keep lacking the one resource that would definitely help it overcome all the present shortcomings that have prevented Europe from finding its true role and mission through the CSDP. Member States remain central to EU security and defence policy. This is why this collection of essays is so valuable for assessing in no uncertain way the long road that lies ahead for any progress to be made. Pierre VIMONT Senior Associate at Carnegie Europe Former Executive Secretary-General of the European External Action Service

Controlo fuzzy semi-activo para aparelhos de amortecimento variável na protecção sísmica

Trabalho apresentado no 10º Congresso Nacional de Sismologia e Engenharia Sísmica, 20-22 abril de 2016, Ponta Delgada, Açores, Portugal

Optimization of the Modular Educational Program Structure

Optimized structure of the educational program consisting of a set of the interconnected educational objects is offered by means of problem solution of optimum partition of the acyclic weighed graph. The condition of acyclicity preservation for subgraphs is formulated and the quantitative assessment of decision options is executed. The original algorithm of search of quasioptimum partition using the genetic algorithm scheme with coding chromosomes by permutation is offered. Object-oriented realization of algorithm in language C++ is described and results of numerical experiments are presented.