A Sparse Bayesian Approach for Joint Feature Selection and Classifier Learning

Peer-reviewed

Efficient Peer Selection in P2P JXTA-Based Platforms

Peer-reviewed

An Experimental Study on Peer Selection in a P2P Network over PlanetLab

Peer-reviewed

Design and evaluation of learning algorithms for dynamic resource management in virtual networks

Network virtualisation is considerably gaining attentionas a solution to ossification of the Internet. However, thesuccess of network virtualisation will depend in part on how efficientlythe virtual networks utilise substrate network resources.In this paper, we propose a machine learning-based approachto virtual network resource management. We propose to modelthe substrate network as a decentralised system and introducea learning algorithm in each substrate node and substrate link,providing self-organization capabilities. We propose a multiagentlearning algorithm that carries out the substrate network resourcemanagement in a coordinated and decentralised way. The taskof these agents is to use evaluative feedback to learn an optimalpolicy so as to dynamically allocate network resources to virtualnodes and links. The agents ensure that while the virtual networkshave the resources they need at any given time, only the requiredresources are reserved for this purpose. Simulations show thatour dynamic approach significantly improves the virtual networkacceptance ratio and the maximum number of accepted virtualnetwork requests at any time while ensuring that virtual networkquality of service requirements such as packet drop rate andvirtual link delay are not affected.

Detection of severe obstructive sleep apnea through voice analysis

tThis paper deals with the potential and limitations of using voice and speech processing to detect Obstruc-tive Sleep Apnea (OSA). An extensive body of voice features has been extracted from patients whopresent various degrees of OSA as well as healthy controls. We analyse the utility of a reduced set offeatures for detecting OSA. We apply various feature selection and reduction schemes (statistical rank-ing, Genetic Algorithms, PCA, LDA) and compare various classifiers (Bayesian Classifiers, kNN, SupportVector Machines, neural networks, Adaboost). S-fold crossvalidation performed on 248 subjects showsthat in the extreme cases (that is, 127 controls and 121 patients with severe OSA) voice alone is able todiscriminate quite well between the presence and absence of OSA. However, this is not the case withmild OSA and healthy snoring patients where voice seems to play a secondary role. We found that thebest classification schemes are achieved using a Genetic Algorithm for feature selection/reduction.

Feature selection for automatic analysis of emotional response based on nonlinear speech modeling suitable for diagnosis of Alzheimer׳s disease

Alzheimer׳s disease (AD) is the most common type of dementia among the elderly. This work is part of a larger study that aims to identify novel technologies and biomarkers or features for the early detection of AD and its degree of severity. The diagnosis is made by analyzing several biomarkers and conducting a variety of tests (although only a post-mortem examination of the patients’ brain tissue is considered to provide definitive confirmation). Non-invasive intelligent diagnosis techniques would be a very valuable diagnostic aid. This paper concerns the Automatic Analysis of Emotional Response (AAER) in spontaneous speech based on classical and new emotional speech features: Emotional Temperature (ET) and fractal dimension (FD). This is a pre-clinical study aiming to validate tests and biomarkers for future diagnostic use. The method has the great advantage of being non-invasive, low cost, and without any side effects. The AAER shows very promising results for the definition of features useful in the early diagnosis of AD.

Sähköasemainvestointihankkeen määrittely

Tämän työn tavoitteena oli tehdä perusteltu esitys kahden suunnitellun sähköaseman toiminnan aloittamisen ajankohdasta sekä rakentamisen kestosta lupamenettelyineen. Työssä oli pyrkimys perustella asemien maantieteellinen sijainti, toteutustapa, rakenne sekä sähköasemien syöttöön tarvittavien 110 kV johtojen alustava rakenne ja reitti. Työssä on tarkasteltu uuden sähköaseman verkosto- ja kustannusvaikutuksia, sähköasemarakenteita ja niiden valintaa, sähköasemainvestointihankkeen vaiheita ja rakennuttamisprosessin läpivientiin tarvittavaa aikaa. Verkoston nykytilaa, käyttövarmuutta ja selviytymistä kuormituksen kasvusta on tarkasteltu sähköasemien toiminnan aloittamisen ajankohtien määrittämiseksi. Työn keskeisin painopiste oli sähköasemien rakentamisajankohdan määrittäminen. Sähköasemien toiminnan aloittamisen ajankohdan määrääväksi tekijäksi muodostuivat sähköasemien korvaustilanteet. Nykytilassa Valkealan haja-asutusalueen sähkönjakelua ei voida taata yksittäisen sähköaseman korvaustilanteessa, minkä takia sähköasemahankkeen valmistelu on aloitettava välittömästi. Kouvolan ydinkeskustan alueen maakaapeliverkon tehonsiirtokyky ja päämuuntajien reservitehokapasiteetti korvaustilanteissa riittävät vielä 5-10 vuotta riippuen suuresti alueen kuormituksen kasvusta.

A hybrid feature selection approach for the early diagnosis of Alzheimer's disease

Objective. Recently, significant advances have been made in the early diagnosis of Alzheimer’s disease from EEG. However, choosing suitable measures is a challenging task. Among other measures, frequency Relative Power and loss of complexity have been used with promising results. In the present study we investigate the early diagnosis of AD using synchrony measures and frequency Relative Power on EEG signals, examining the changes found in different frequency ranges. Approach. We first explore the use of a single feature for computing the classification rate, looking for the best frequency range. Then, we present a multiple feature classification system that outperforms all previous results using a feature selection strategy. These two approaches are tested in two different databases, one containing MCI and healthy subjects (patients age: 71.9 ± 10.2, healthy subjects age: 71.7 ± 8.3), and the other containing Mild AD and healthy subjects (patients age: 77.6 ± 10.0; healthy subjects age: 69.4± 11.5). Main Results. Using a single feature to compute classification rates we achieve a performance of 78.33% for the MCI data set and of 97.56 % for Mild AD. Results are clearly improved using the multiple feature classification, where a classification rate of 95% is found for the MCI data set using 11 features, and 100% for the Mild AD data set using 4 features. Significance. The new features selection method described in this work may be a reliable tool that could help to design a realistic system that does not require prior knowledge of a patient's status. With that aim, we explore the standardization of features for MCI and Mild AD data sets with promising results.

Learning Preferences with Kernel-Based Methods

Learning of preference relations has recently received significant attention in machine learning community. It is closely related to the classification and regression analysis and can be reduced to these tasks. However, preference learning involves prediction of ordering of the data points rather than prediction of a single numerical value as in case of regression or a class label as in case of classification. Therefore, studying preference relations within a separate framework facilitates not only better theoretical understanding of the problem, but also motivates development of the efficient algorithms for the task. Preference learning has many applications in domains such as information retrieval, bioinformatics, natural language processing, etc. For example, algorithms that learn to rank are frequently used in search engines for ordering documents retrieved by the query. Preference learning methods have been also applied to collaborative filtering problems for predicting individual customer choices from the vast amount of user generated feedback. In this thesis we propose several algorithms for learning preference relations. These algorithms stem from well founded and robust class of regularized least-squares methods and have many attractive computational properties. In order to improve the performance of our methods, we introduce several non-linear kernel functions. Thus, contribution of this thesis is twofold: kernel functions for structured data that are used to take advantage of various non-vectorial data representations and the preference learning algorithms that are suitable for different tasks, namely efficient learning of preference relations, learning with large amount of training data, and semi-supervised preference learning. Proposed kernel-based algorithms and kernels are applied to the parse ranking task in natural language processing, document ranking in information retrieval, and remote homology detection in bioinformatics domain. Training of kernel-based ranking algorithms can be infeasible when the size of the training set is large. This problem is addressed by proposing a preference learning algorithm whose computation complexity scales linearly with the number of training data points. We also introduce sparse approximation of the algorithm that can be efficiently trained with large amount of data. For situations when small amount of labeled data but a large amount of unlabeled data is available, we propose a co-regularized preference learning algorithm. To conclude, the methods presented in this thesis address not only the problem of the efficient training of the algorithms but also fast regularization parameter selection, multiple output prediction, and cross-validation. Furthermore, proposed algorithms lead to notably better performance in many preference learning tasks considered.

Defining housing market areas using commuting and migration algorithms. Catalonia (Spain) as an applied case study

In the literature on housing market areas, different approaches can be found to defining them, for example, using travel-to-work areas and, more recently, making use of migration data. Here we propose a simple exercise to shed light on which approach performs better. Using regional data from Catalonia, Spain, we have computed housing market areas with both commuting data and migration data. In order to decide which procedure shows superior performance, we have looked at uniformity of prices within areas. The main finding is that commuting algorithms present more homogeneous areas in terms of housing prices.

Threshold Selection Criteria for Quantification of Lumbosacral Cerebrospinal Fluid and Root Volumes from MRI

BACKGROUND AND PURPOSE: The high variability of CSF volumes partly explains the inconsistency of anesthetic effects, but may also be due to image analysis itself. In this study, criteria for threshold selection are anatomically defined. METHODS: T2 MR images (n = 7 cases) were analyzed using 3-dimentional software. Maximal-minimal thresholds were selected in standardized blocks of 50 slices of the dural sac ending caudally at the L5-S1 intervertebral space (caudal blocks) and middle L3 (rostral blocks). Maximal CSF thresholds: threshold value was increased until at least one voxel in a CSF area appeared unlabeled and decreased until that voxel was labeled again: this final threshold was selected. Minimal root thresholds: thresholds values that selected cauda equina root area but not adjacent gray voxels in the CSF-root interface were chosen. RESULTS: Significant differences were found between caudal and rostral thresholds. No significant differences were found between expert and nonexpert observers. Average max/min thresholds were around 1.30 but max/min CSF volumes were around 1.15. Great interindividual CSF volume variability was detected (max/min volumes 1.6-2.7). CONCLUSIONS: The estimation of a close range of CSF volumes which probably contains the real CSF volume value can be standardized and calculated prior to certain intrathecal procedures

Learning and selection processes

In this paper I defend a teleological explanation of normativity, i. e., I argue that what an organism (or device) is supposed to do is determined by its etiological function. In particular, I present a teleological account of the normativity that arises in learning processes, and I defend it from some objections

Construction and validation of a scoring system for the selection of high-quality data in a Spanish population primary care database (SIDIAP)

Background Computerised databases of primary care clinical records are widely used for epidemiological research. In Catalonia, the InformationSystem for the Development of Research in Primary Care (SIDIAP) aims to promote the development of research based on high-quality validated data from primary care electronic medical records. Objective The purpose of this study is to create and validate a scoring system (Registry Quality Score, RQS) that will enable all primary care practices (PCPs) to be selected as providers of researchusable data based on the completeness of their registers. Methods Diseases that were likely to be representative of common diagnoses seen in primary care were selected for RQS calculations. The observed/ expected cases ratio was calculated for each disease. Once we had obtained an estimated value for this ratio for each of the selected conditions we added up the ratios calculated for each condition to obtain a final RQS. Rate comparisons between observed and published prevalences of diseases not included in the RQS calculations (atrial fibrillation, diabetes, obesity, schizophrenia, stroke, urinary incontinenceand Crohn’s disease) were used to set the RQS cutoff which will enable researchers to select PCPs with research-usable data. Results Apart from Crohn’s disease, all prevalences were the same as those published from the RQS fourth quintile (60th percentile) onwards. This RQS cut-off provided a total population of 1 936 443 (39.6% of the total SIDIAP population). Conclusions SIDIAP is highly representative of the population of Catalonia in terms of geographical, age and sex distributions. We report the usefulness of rate comparison as a valid method to establish research-usable data within primary care electronic medical records

Citation auctions as a method to improve selection of scientific papers

This paper describes the basis of citation auctions as a new approach to selecting scientific papers for publication. Our main idea is to use an auction for selecting papers for publication through - differently from the state of the art - bids that consist of the number of citations that a scientist expects to receive if the paper is published. Hence, a citation auction is the selection process itself, and no reviewers are involved. The benefits of the proposed approach are two-fold. First, the cost of refereeing will be either totally eliminated or significantly reduced, because the process of citation auction does not need prior understanding of the paper's content to judge the quality of its contribution. Additionally, the method will not prejudge the content of the paper, so it will increase the openness of publications to new ideas. Second, scientists will be much more committed to the quality of their papers, paying close attention to distributing and explaining their papers in detail to maximize the number of citations that the paper receives. Sample analyses of the number of citations collected in papers published in years 1999-2004 for one journal, and in years 2003-2005 for a series of conferences (in a totally different discipline), via Google scholar, are provided. Finally, a simple simulation of an auction is given to outline the behaviour of the citation auction approach

Horizontal Grid Size Selection and its Influence on Mesoscale Model Simulations

The use of two-dimensional spectral analysis applied to terrain heights in order to determine characteristic terrain spatial scales and its subsequent use for the objective definition of an adequate grid size required to resolve terrain forcing are presented in this paper. In order to illustrate the influence of grid size, atmospheric flow in a complex terrain area of the Spanish east coast is simulated by the Regional Atmospheric Modeling System (RAMS) mesoscale numerical model using different horizontal grid resolutions. In this area, a grid size of 2 km is required to account for 95% of terrain variance. Comparison among results of the different simulations shows that, although the main wind behavior does not change dramatically, some small-scale features appear when using a resolution of 2 km or finer. Horizontal flow pattern differences are significant both in the nighttime, when terrain forcing is more relevant, and in the daytime, when thermal forcing is dominant. Vertical structures also are investigated, and results show that vertical advection is influenced highly by the horizontal grid size during the daytime period. The turbulent kinetic energy and potential temperature vertical cross sections show substantial differences in the structure of the planetary boundary layer for each model configuration