Learning Recursive Bayesian Multinets for Data Clustering by Means of Constructive Induction

This paper introduces and evaluates a new class of knowledge model, the recursive Bayesian multinet (RBMN), which encodes the joint probability distribution of a given database. RBMNs extend Bayesian networks (BNs) as well as partitional clustering systems. Briefly, a RBMN is a decision tree with component BNs at the leaves. A RBMN is learnt using a greedy, heuristic approach akin to that used by many supervised decision tree learners, but where BNs are learnt at leaves using constructive induction. A key idea is to treat expected data as real data. This allows us to complete the database and to take advantage of a closed form for the marginal likelihood of the expected complete data that factorizes into separate marginal likelihoods for each family (a node and its parents). Our approach is evaluated on synthetic and real-world databases.     

A hybrid dynamic motion prediction method for multibody digital human models based on a motion database and motion knowledge

In this paper, we present a novel method to predict human motion, seeking to combine the advantages of both data-based and knowledge-based motion prediction methods. Our method relies on a database of captured motions for reference and introduces knowledge in the prediction in the form of a motion control law, which is followed while resembling the actually performed reference motion. The prediction is carried out by solving an optimization problem in which the following conditions are imposed to the motion: must fulfill the goals of the task; resemble the reference motion selected from the database; follow a knowledge-based dynamic motion control law; and ensure the dynamic equilibrium of the human model, considering its interactions with the environment. In this work, we apply the proposed method to a database of clutch pedal depression motions, and we present the results for three predictions. The method is validated by comparing the results of the prediction to motions actually performed in similar conditions. The predicted motions closely resemble the motions in the validation database and no significant differences have been noted either in the motion’s kinematics or in the motion’s dynamics.     

Regularized continuous estimation of distribution algorithms

Regularization is a well-known technique in statistics for model estimation which is used to improve the generalization ability of the estimated model. Some of the regularization methods can also be used for variable selection that is especially useful in high-dimensional problems. This paper studies the use of regularized model learning in estimation of distribution algorithms (EDAs) for continuous optimization based on Gaussian distributions. We introduce two approaches to the regularized model estimation and analyze their effect on the accuracy and computational complexity of model learning in EDAs. We then apply the proposed algorithms to a number of continuous optimization functions and compare their results with other Gaussian distribution-based EDAs. The results show that the optimization performance of the proposed RegEDAs is less affected by the increase in the problem size than other EDAs, and they are able to obtain significantly better optimization values for many of the functions in high-dimensional settings.     

Detecting reliable gene interactions by a hierarchy of Bayesian network classifiers

The main purpose of a gene interaction network is to map the relationships of the genes that are out of sight when a genomic study is tackled. DNA microarrays allow the measure of gene expression of thousands of genes at the same time. These data constitute the numeric seed for the induction of the gene networks. In this paper, we propose a new approach to build gene networks by means of Bayesian classifiers, variable selection and bootstrap resampling. The interactions induced by the Bayesian classifiers are based both on the expression levels and on the phenotype information of the supervised variable. Feature selection and bootstrap resampling add reliability and robustness to the overall process removing the false positive findings. The consensus among all the induced models produces a hierarchy of dependences and, thus, of variables. Biologists can define the depth level of the model hierarchy so the set of interactions and genes involved can vary from a sparse to a dense set. Experimental results show how these networks perform well on classification tasks. The biological validation matches previous biological findings and opens new hypothesis for future studies.     

Imhotep: an approach to user and device conscious mobile applications

As the dependence on mobile devices increases, the need for supporting a wider range of users and devices becomes crucial. Elders and people with disabilities adopt new technologies reluctantly, a tendency caused by the lack of adaptation of these technologies to their needs. To address this challenge, this paper describes a framework, Imhotep, whose aim is to aid developers in the accessible application creation process, making the creation of user-centered applications easier and faster. Our framework allows to easily adapt the applications to the constraints imposed by the user capabilities (sensorial, cognitive, and physical capabilities) and device capabilities by providing a repository that will manage the compilation and deployment of applications that include a set of preprocessor directives in the source code. These directives are enhanced with concepts that are automatically adjusted to the current trends of mobile devices by using a Fuzzy Knowledge-Eliciting Reasoner. Our final goal is to increase the number of applications targeted to elders and people with disabilities providing tools that facilitate their development. The paper also describes the evaluation of both the accuracy of the fuzzy terms generated for mobile devices and the usability of the proposed platform.     

Advances and Perspectives in Dental Pulp Stem Cell Based Neuroregeneration Therapies

Human dental pulp stem cells (hDPSCs) are some of the most promising stem cell types for regenerative therapies given their ability to grow in the absence of serum and their realistic possibility to be used in autologous grafts. In this review, we describe the particular advantages of hDPSCs for neuroregenerative cell therapies. We thoroughly discuss the knowledge about their embryonic origin and characteristics of their postnatal niche, as well as the current status of cell culture protocols to maximize their multilineage differentiation potential, highlighting some common issues when assessing neuronal differentiation fates of hDPSCs. We also review the recent progress on neuroprotective and immunomodulatory capacity of hDPSCs and their secreted extracellular vesicles, as well as their combination with scaffold materials to improve their functional integration on the injured central nervous system (CNS) and peripheral nervous system (PNS). Finally, we offer some perspectives on the current and possible future applications of hDPSCs in neuroregenerative cell therapies.     

Cooperative Distributed MIMO Channels in Wireless Sensor Networks

The large number of network nodes and the energy constraints make Wireless Sensor Networks (WSN) one of the most important application fields for Cooperative Diversity. Node cooperation increases the spatial diversity of wireless channels and, thus, reduces the transmitted power. In this paper, we propose a multi-hop WSN with nodes grouped in cooperative clusters that exploits transmit and receive cooperation among cluster nodes. Multi-hop transmission is carried out by concatenating single cluster-to-cluster hops, where every cluster-to-cluster link is defined as a cooperative distributed multiple-input-multiple-output (MIMO) channel. Transmit diversity is exploited through a time-division, decoder-and-forward, relaying scheme based upon two time slots: the Intracluster Slot, used for data sharing within the cluster, and the Intercluster Slot, used for transmission between clusters. At the receiver side, a distributed reception protocol is devised based upon a Selection Diversity algorithm. The proposed multi-hop cooperative WSN is optimally designed for minimum end-to-end outage probability by deriving the optimum time and power allocated on the intracluster and intercluster slots of every single hop, given a per-link energy constraint. A simplified suboptimum resource allocation is also proposed, which performs close to the optimal policy. Results show that the proposed scheme achieves diversity equal to the equivalent MIMO system and significantly reduces energy consumption with respect to. the non-cooperative channel     

Isotropic Image Analysis for Improving CBR Forecasting

A novel hybrid forecasting Case-Based Reasoning (CBR) system is presented in this interdisciplinary study in which an isotropic buffer operator is applied for case-based creation. Commonly used as an image analysis technique by commercial Geographic Information Systems (GIS), the buffer operator in this particular system calculates the area of an oil slick for prediction and visualization tasks. The use of the buffer operator improves the quality of the data used by the system and in consequence the quality of the results obtained. The system generates predictions by using historical data on oil-slick formation following a spill.