Larranaga P.,Computational Intelligence Group |
Karshenas H.,Computational Intelligence Group |
Bielza C.,Computational Intelligence Group |
Santana R.,University of the Basque Country
Information Sciences | Year: 2013
Thanks to their inherent properties, probabilistic graphical models are one of the prime candidates for machine learning and decision making tasks especially in uncertain domains. Their capabilities, like representation, inference and learning, if used effectively, can greatly help to build intelligent systems that are able to act accordingly in different problem domains. Bayesian networks are one of the most widely used class of these models. Some of the inference and learning tasks in Bayesian networks involve complex optimization problems that require the use of meta-heuristic algorithms. Evolutionary algorithms, as successful problem solvers, are promising candidates for this purpose. This paper reviews the application of evolutionary algorithms for solving some NP-hard optimization tasks in Bayesian network inference and learning. © 2013 Elsevier Inc. All rights reserved.
News Article | September 9, 2016
The thesis, titled "Computational intelligent methods for trusting in social networks," produced by the computer engineer David Núñez in the Computational Intelligence Group at the UPV/EHU's Faculty of Computing, falls within the framework of the European research project Social and Smart (SandS). A part of the project focuses its attention on user interaction with smart domestic appliances linked to a smart module. These are household appliances (systems) to which the user describes in ordinary language the problem that he/she wants to solve (such as "making bread," "removing a stain from a pair of trousers," etc. depending on the type of household appliance). The system analyses the problem that needs to be solved and searches the database to see whether there is a solution (recipe) for the problem described by the user. If one exists, it is provided, and if not, the system forwards the description of the problem to an intelligent module so that a new solution can be produced and then passed on to the user. The user can execute the proposed solution or else readjust its parameters. Once the execution of the problem has been completed, the user will express his/her satisfaction with the result obtained. The users can communicate with each other over the system's social network and propose recipes that can be evaluated by other users. The thesis by Núñez has provided new intelligent techniques in the area of social networks. Specifically, he has covered three lines of research in this area: trust, the recommendation systems and the maximising of influence. The first line of research seeks to predict the trust that a user will place in another person belonging to her social environment on the basis of the opinions that other contacts have expressed about the target user. The researcher has developed some tools for predicting trust that are more straightforward than the ones found in the literature, and are more algebra-based. The second line of research focuses on the systems of recommendation, and two experiments have been carried out. The first is linked to the generating of recipes for making bread in a smart bread maker. An attempt has been made to simulate the prediction of the bread recipe (solution of the problem) on the basis of the satisfaction expressed (description of the problem), and even to predict satisfaction (solution of the problem) on the basis of a recipe provided (description of the problem). The second task in this second line of research has endeavoured to make recommendations about products. The recommendation is based on the previous evaluations of the users. What is being proposed are techniques based on the Web of Trust of the target user to whom one wishes to make a recommendation and also on similarities between users and their means of evaluation. The third line of research is related to maximising influence. The aim of this line is to detect what would be the minimum set of users of a social network that is capable of influencing the maximum possible number of users of the network. "We have come up with a new algorithm that improves the algorithm that exists in the literature in terms of time—the classical Greedy method," explained David Núñez. "Our method has succeeded in getting closer to the optimum like the Greedy one, but does so more rapidly." Explore further: Who goes there? Verifying identity online More information: J. David Nuñez-Gonzalez et al, A new heuristic for influence maximization in social networks, Logic Journal of IGPL (2016). DOI: 10.1093/jigpal/jzw048
Savio A.,Computational Intelligence Group |
Grana M.,Computational Intelligence Group
Expert Systems with Applications | Year: 2013
Deformation-based Morphometry (DBM) allows detection of significant morphological differences of brain anatomy, such as those related to brain atrophy in Alzheimer's Disease (AD). DBM process is as follows: First, performs the non-linear registration of a subject's structural MRI volume to a reference template. Second, computes scalar measures of the registration's deformation field. Third, performs across volume statistical group analysis of these scalar measures to detect effects. In this paper we use the scalar deformation measures for Computer Aided Diagnosis (CAD) systems for AD. Specifically this paper deals with feature extraction methods over five such scalar measures. We evaluate three supervised feature selection methods based on voxel site significance measures given by Pearson correlation, Bhattacharyya distance and Welch's t-test, respectively. The CAD system discriminating between healthy control subjects (HC) and AD patients consists of a Support Vector Machine (SVM) classifier trained on the DBM selected features. The paper reports experimental results on structural MRI data from the cross-sectional OASIS database. Average 10-fold cross-validation classification results are comparable or improve the state-of-The-Art results of other approaches performing CAD from structural MRI data. Localization in the brain of the most discriminant deformation voxel sites is in agreement with findings reported in the literature. © 2012 Elsevier Ltd. All rights reserved.
Grana M.,Computational Intelligence Group |
Savio A.M.,Computational Intelligence Group |
Garcia-Sebastian M.,Computational Intelligence Group |
Fernandez E.,Computational Intelligence Group
Image and Vision Computing | Year: 2010
We introduce an approach to fMRI analysis based on the Endmember Induction Heuristic Algorithm (EIHA). This algorithm uses the Lattice Associative Memory (LAM) to detect Lattice Independent vectors, which can be assumed to be Affine Independent, and therefore candidates to be the endmembers of the data. Induced endmembers are used to compute the activation levels of voxels as result of an unmixing process. The endmembers correspond to diverse activation patterns, one of these activation patterns corresponds to the resting state of the neuronal tissue. The on-line working of the algorithm does not need neither a previous training process nor a priori models of the data. Results on a case study compare with the results given by the state of art SPM software. © 2009 Elsevier B.V. All rights reserved.