Search

Abstract: The use of binary support vector machines (SVMs) in multi-classification is addressed in this paper. Margins associated to the bi-classifiers, since they depend on the geometrical disposition of the classes being separated, are, in general, of various magnitudes. In order to overcome this scaling problem, a normalization process should be applied on the SVMs’ outputs. Thus, a new normalization approach is presented based on the convex hulls that contain the classes to be separated. Furthermore, a theoretical study is developed which justifies the proposed approach, and an interpretation is provided. An empirical study is also carried out to compare this normalization with others found in the literature. [Copyright &y& Elsevier]

Abstract: Energy saving involves two direct benefits: sustainability and cost reduction. Within the field of Information Technology, clusters, grids and data centres represent the hungriest consumers of energy and therefore energy (saving) policies for these infrastructures should be applied in order to maximize their resources. It is proved in this paper that approximately 40% of energy can be saved in a data centre if an adequate policy is applied. Furthermore, a software tool is presented where simulations can be run and results for real scenarios can be obtained. [Copyright &y& Elsevier]

Abstract: This paper introduces a new method to implement a motion recognition process using a mobile phone fitted with an accelerometer. The data collected from the accelerometer are interpreted by means of a statistical study and machine learning algorithms in order to obtain a classification function. Then, that function is implemented in a mobile phone and online experiments are carried out. Experimental results show that this approach can be used to effectively recognize different human activities with a high-level accuracy. [Copyright &y& Elsevier]

Abstract: In this paper, a generalization of support vector machines is explored where it is considered that input vectors have different norms for each class. It is proved that the optimization problem for binary classification by using the maximal margin principle with and norms only depends on the norm if . Furthermore, the selection of a different bias in the classifier function is a consequence of the norm in this approach. Some commentaries on the most commonly used approaches of SVM are also given as particular cases. [Copyright &y& Elsevier]

In this paper the k-nearest-neighbours (KNN) based method is presented for the classification of time series which use qualitative learning to identify similarities using kernels. To this end, time series are transformed into symbol strings by means of several discretization methods and a distance based on a kernel between symbols in ordinal scale is used to calculate the similarity between time series. Hence, the idea proposed is the consideration of the simultaneous use of symbolic representation together with a kernel based approach for classification of time series. The methodology has been tested and compared with quantitative learning from a television-viewing shared data set and has yielded a high success identification ratio. [Copyright &y& Elsevier]

This paper describes a new discretization algorithm, called Ameva, which is designed to work with supervised learning algorithms. Ameva maximizes a contingency coefficient based on Chi-square statistics and generates a potentially minimal number of discrete intervals. Its most important advantage, in contrast with several existing discretization algorithms, is that it does not need the user to indicate the number of intervals. We have compared Ameva with one of the most relevant discretization algorithms, CAIM. Tests performed comparing these two algorithms show that discrete attributes generated by the Ameva algorithm always have the lowest number of intervals, and even if the number of classes is high, the same computational complexity is maintained. A comparison between the Ameva and the genetic algorithm approaches has been also realized and there are very small differences between these iterative and combinatorial approaches, except when considering the execution time. [Copyright &y& Elsevier]

A person's preference to select or reject certain meals is influenced by several aspects, including colour. In this paper, we study the relevance of food colour for such preferences. To this end, a set of images of meals is processed by an automatic method that associates mood adjectives that capture such meal preferences. These adjectives are obtained by analyzing the colour palettes in the image, using a method based in Kobayashi's model of harmonic colour combinations. The paper also validates that the colour palettes calculated for each image are harmonic by developing a rating model to predict how much a user would like the colour palettes obtained. This rating is computed using a regression model based on the COLOURlovers dataset implemented to learn users' preferences. Finally, the adjectives associated automatically with images of dishes are validated by a survey which was responded by 178 people and demonstrates that the labels are adequate. The results obtained in this paper have applications in tourism marketing, to help in the design of marketing multimedia material, especially for promoting restaurants and gastronomic destinations. [ABSTRACT FROM AUTHOR]

Abstract: Human activity recognition systems are currently implemented by hundreds of applications and, in recent years, several technology manufacturers have introduced new wearable devices for this purpose. Battery consumption constitutes a critical point in these systems since most are provided with a rechargeable battery. In this paper, by using discrete techniques based on the Ameva algorithm, an innovative approach for human activity recognition systems on mobile devices is presented. Furthermore, unlike other systems in current use, this proposal enables recognition of high granularity activities by using accelerometer sensors. Hence, the accuracy of activity recognition systems can be increased without sacrificing efficiency. A comparative is carried out between the proposed approach and an approach based on the well-known neural networks. [Copyright &y& Elsevier]

Abstract: In this paper, a system based on the generation of a Hidden Markov Model from the past GPS log and current location is presented to predict a user’s destination when beginning a new trip. This approach drastically reduces the number of points supplied by the GPS device and it permits a “support-map” to be generated in which the main characteristics of the trips for each user are taken into account. Hence, in contrast with other similar approaches, total independence from a street-map database is achieved. [ABSTRACT FROM AUTHOR]

Abstract: The use of data represented by intervals can be caused by imprecision in the input information, incompleteness in patterns, discretization procedures, prior knowledge insertion or speed-up learning. All the existing support vector machine (SVM) approaches working on interval data use local kernels based on a certain distance between intervals, either by combining the interval distance with a kernel or by explicitly defining an interval kernel. This article introduces a new procedure for the linearly separable case, derived from convex optimization theory, inserting information directly into the standard SVM in the form of intervals, without taking any particular distance into consideration. [Copyright &y& Elsevier]

Abstract: The recent boom of GPS (Global Positioning System) as a universal method of location has meant that most people in developed countries have already used this technology sometime in their lives. However, this system suffers from an ever-increasing problem: energy expenditure. GPS receivers have been integrated into increasingly smaller devices such as the latest generation of mobiles, thereby making battery-saving a priority in the use of this technology. This article lays out a series of ideas which, through the use of auxiliary technologies, are able to maximize energy saving. By means of outdoor exit detection, it will be possible to automatically disconnect the GPS while the user stays indoors and later reconnect it on leaving the building. [Copyright &y& Elsevier]

Abstract: A novel video similarity measure is proposed by using visual features, alignment distances and speech transcripts. First, video files are represented by a sequence of segments each of which contains colour histograms, starting time, and a set of phonemes. After, textual, alignment and visual features are extracted of these segments. The following step, bipartite matching and statistical features are applied to find correspondences between segments. Finally, a similarity is calculated between videos. Experiments have been carried out and promising results have been obtained. [Copyright &y& Elsevier]