Short-term Feature Space and Music Genre Classification

In music genre classification, most approaches rely on statistical characteristics of low-level features computed on short audio frames. In these methods, it is implicitly considered that frames carry equally relevant information loads and that either individual frames, or distributions thereof, somehow capture the specificities of each genre. In this paper we study the representation space defined by short-term audio features with respect to class boundaries, and compare different processing techniques to partition this space. These partitions are evaluated in terms of accuracy on two genre classification tasks, with several types of classifiers. Experiments show that a randomized and unsupervised partition of the space, used in conjunction with a Markov Model classifier lead to accuracies comparable to the state of the art. We also show that unsupervised partitions of the space tend to create less hubs.

TBCCD1, a new centrosomal protein, is required for centrosome and Golgi apparatus positioning

In animal cells the centrosome is positioned at the cell centre in close association with the nucleus. The mechanisms responsible for this are not completely understood. Here, we report the first characterization of human TBCC-domain containing 1 (TBCCD1), a protein related to tubulin cofactor C. TBCCD1 localizes at the centrosome and at the spindle midzone, midbody and basal bodies of primary and motile cilia. Knockdown of TBCCD1 in RPE-1 cells caused the dissociation of the centrosome from the nucleus and disorganization of the Golgi apparatus. TBCCD1-depleted cells are larger, less efficient in primary cilia assembly and their migration is slower in wound-healing assays. However, the major microtubule-nucleating activity of the centrosome is not affected by TBCCD1 silencing. We propose that TBCCD1 is a key regulator of centrosome positioning and consequently of internal cell organization.

Synthesis, characterization, electrochemical behavior and in vitro protein tyrosine kinase inhibitory activity of the cymene-halogenobenzohydroxamato [Ru(eta(6)-cymene)(bha)Cl] complexes

The ruthenium(II)-cymene complexes [Ru(eta(6)-cymene)(bha)Cl] with substituted halogenobenzohydroxamato (bha) ligands (substituents = 4-F, 4-Cl, 4-Br, 2,4-F-2, 3,4-F-2, 2,5-F-2, 2,6-F-2) have been synthesized and characterized by elemental analysis, IR, H-1 NMR, C-13 NMR, cyclic voltammetry and controlled-potential electrolysis, and density functional theory (DFT) studies. The compositions of their frontier molecular orbitals (MOs) were established by DFT calculations, and the oxidation and reduction potentials are shown to follow the orders of the estimated vertical ionization potential and electron affinity, respectively. The electrochemical E-L Lever parameter is estimated for the first time for the various bha ligands, which can thus be ordered according to their electron-donor character. All complexes exhibit very strong protein tyrosine kinase (PTK) inhibitory activity, even much higher than that of genistein, the clinically used PTK inhibitory drug. The complex containing the 2,4-difluorobenzohydroxamato ligand is the most active one, and the dependences of the PTK activity of the complexes and of their redox potentials on the ring substituents are discussed. (C) 2012 Elsevier B.V. All rights reserved.

Data mining framework for fatty liver disease classification in ultrasound: a hybrid feature extraction paradigm

PURPOSE: Fatty liver disease (FLD) is an increasing prevalent disease that can be reversed if detected early. Ultrasound is the safest and ubiquitous method for identifying FLD. Since expert sonographers are required to accurately interpret the liver ultrasound images, lack of the same will result in interobserver variability. For more objective interpretation, high accuracy, and quick second opinions, computer aided diagnostic (CAD) techniques may be exploited. The purpose of this work is to develop one such CAD technique for accurate classification of normal livers and abnormal livers affected by FLD. METHODS: In this paper, the authors present a CAD technique (called Symtosis) that uses a novel combination of significant features based on the texture, wavelet transform, and higher order spectra of the liver ultrasound images in various supervised learning-based classifiers in order to determine parameters that classify normal and FLD-affected abnormal livers. RESULTS: On evaluating the proposed technique on a database of 58 abnormal and 42 normal liver ultrasound images, the authors were able to achieve a high classification accuracy of 93.3% using the decision tree classifier. CONCLUSIONS: This high accuracy added to the completely automated classification procedure makes the authors' proposed technique highly suitable for clinical deployment and usage.

Translation termination and protein folding pathway genes are not correlated in gastric cancer

Background: The eukaryotic release factor 3 (eRF3) has been shown to affect both tubulin and actin cytoskeleton, suggesting a role in cytoskeleton assembly, mitotic spindle formation and chromosome segregation. Also, direct interactions between eRF3 and subunits of the cytosolic chaperonin CCT have been described. Moreover, both eRF3a and CCT subunits have been described to be up-regulated in cancer tissues. Our aim was to evaluate the hypothesis that eRF3 expression levels are correlated with the expression of genes encoding proteins involved in the tubulin folding pathways. Methods: Relative expression levels of eRF1, eRF3a/GSPT1, PFDN4, CCT2, CCT4, and TBCA genes in tumour samples relative to their adjacent normal tissues were investigated using real time-polymerase chain reaction in 20 gastric cancer patients. Results: The expression levels of eRF3a/GSPT1 were not correlated with the expression levels of the other genes studied. However, significant correlations were detected between the other genes, both within intestinal and diffuse type tumours. Conclusions: eRF3a/GSPT1 expression at the mRNA level is independent from both cell translation rates and from the expression of the genes involved in tubulin-folding pathways. The differences in the patterns of expression of the genes studied support the hypothesis of genetically independent pathways in the origin of intestinal and diffuse type gastric tumours.

Besnoitia besnoiti protein disulfide isomerase (BbPDI): molecular characterization, expression and in silico modelling

Besnoitia besnoiti is an apicomplexan parasite responsible for bovine besnoitiosis, a disease with a high prevalence in tropical and subtropical regions and re-emerging in Europe. Despite the great economical losses associated with besnoitiosis, this disease has been underestimated and poorly studied, and neither an effective therapy nor an efficacious vaccine is available. Protein disulfide isomerase (PDI) is an essential enzyme for the acquisition of the correct three-dimensional structure of proteins. Current evidence suggests that in Neosporacaninum and Toxoplasmagondii, which are closely related to B. besnoiti, PDI play an important role in host cell invasion, is a relevant target for the host immune response, and represents a promising drug target and/or vaccine candidate. In this work, we present the nucleotide sequence of the B. besnoiti PDI gene. BbPDI belongs to the thioredoxin-like superfamily (cluster 00388) and is included in the PDI_a family (cluster defined cd02961) and the PDI_a_PDI_a'_c subfamily (cd02995). A 3D theoretical model was built by comparative homology using Swiss-Model server, using as a template the crystallographic deduced model of Tapasin-ERp57 (PDB code 3F8U chain C). Analysis of the phylogenetic tree for PDI within the phylum apicomplexa reinforces the close relationship among B. besnoiti, N. caninum and T. gondii. When subjected to a PDI-assay based on the polymerisation of reduced insulin, recombinant BbPDI expressed in E. coli exhibited enzymatic activity, which was inhibited by bacitracin. Antiserum directed against recombinant BbPDI reacted with PDI in Western blots and by immunofluorescence with B. besnoiti tachyzoites and bradyzoites.

Effect of a sardine supplement on C-reactive protein in patients receiving hemodialysis

Objective - The study evaluated the effect of a canned sardine supplement in C-reactive protein (CRP) in patients on hemodialysis (HD) and the compliance and adherence to this supplement. Design - This was a quasi-experimental study: Participants with a serum CRP of 5 mg/dL or less volunteered to consume a sardine supplement or were maintained on the usual cheese/ham sandwich supplement. Setting - The study took place in two outpatient dialysis units in Lisbon, Portugal. Patients - The study comprised 63 patients receiving maintenance HD three times per week for at least 6 months and an initial CRP concentration of 5 mg/dL or less. Exclusion criteria included the presence of graft vascular access or history of cancer. Intervention - After a 4-week washout period, the nutritional intervention included a canned sardine sandwich for the case group (n = 31) and a cheese or ham sandwich for the control group (n = 32), to be ingested during each routine HD session, 3 times per week, for 8 weeks. Main outcome measure - Serum levels of high-sensitivity CRP were the outcome measure. Results - Only 65 patients from the invited 186 patients met the inclusion criteria and agreed to eat the sardine sandwich supplement three times per week and were involved in the study. A significant proportion of 48% (n = 31, case group) consumed the sardine sandwich supplement three times per week for 8 weeks, fulfilling the requirements and completing the study. The present investigation showed that a sardine sandwich supplement had no effect on CRP levels among patients on HD. However, when participants were stratified according to tertiles of CRP distribution values at baseline, a reduction in CRP levels was found for those in the higher tertile, being higher for the case group (P = .047). Although diabetic patients were excluded from the analysis (eight in the sardine supplementation group and seven in the control group) a significant CRP reduction was found (P = .034). Conclusion - Although a supplement of low-dose n-3 long-chain polyunsaturated fatty acids had no effect on the plasma high-sensitivity CRP of the supplemented group, a reduction in CRP levels was found when patients were stratified for tertiles of CRP (for the upper tertile) and diabetic status (for nondiabetic patients). These findings need to be further confirmed. This canned sardine supplement was accepted by an important proportion of patients, enhancing diet variety and contributing for a greater n-3 long-chain polyunsaturated fatty acids eicosapentaenoic acid and docosahexaenoic acid intake.

An Evaluation of Local Feature Combiners for Robot Visual Localization

In the last decade, local image features have been widely used in robot visual localization. To assess image similarity, a strategy exploiting these features compares raw descriptors extracted from the current image to those in the models of places. This paper addresses the ensuing step in this process, where a combining function must be used to aggregate results and assign each place a score. Casting the problem in the multiple classifier systems framework, we compare several candidate combiners with respect to their performance in the visual localization task. A deeper insight into the potential of the sum and product combiners is provided by testing two extensions of these algebraic rules: threshold and weighted modifications. In addition, a voting method, previously used in robot visual localization, is assessed. All combiners are tested on a visual localization task, carried out on a public dataset. It is experimentally demonstrated that the sum rule extensions globally achieve the best performance. The voting method, whilst competitive to the algebraic rules in their standard form, is shown to be outperformed by both their modified versions.

Feature selection for clustering categorical data with an embedded modelling approach

Research on the problem of feature selection for clustering continues to develop. This is a challenging task, mainly due to the absence of class labels to guide the search for relevant features. Categorical feature selection for clustering has rarely been addressed in the literature, with most of the proposed approaches having focused on numerical data. In this work, we propose an approach to simultaneously cluster categorical data and select a subset of relevant features. Our approach is based on a modification of a finite mixture model (of multinomial distributions), where a set of latent variables indicate the relevance of each feature. To estimate the model parameters, we implement a variant of the expectation-maximization algorithm that simultaneously selects the subset of relevant features, using a minimum message length criterion. The proposed approach compares favourably with two baseline methods: a filter based on an entropy measure and a wrapper based on mutual information. The results obtained on synthetic data illustrate the ability of the proposed expectation-maximization method to recover ground truth. An application to real data, referred to official statistics, shows its usefulness.

Enhancing Multimodal Silent Speech Interfaces With Feature Selection

In research on Silent Speech Interfaces (SSI), different sources of information (modalities) have been combined, aiming at obtaining better performance than the individual modalities. However, when combining these modalities, the dimensionality of the feature space rapidly increases, yielding the well-known "curse of dimensionality". As a consequence, in order to extract useful information from this data, one has to resort to feature selection (FS) techniques to lower the dimensionality of the learning space. In this paper, we assess the impact of FS techniques for silent speech data, in a dataset with 4 non-invasive and promising modalities, namely: video, depth, ultrasonic Doppler sensing, and surface electromyography. We consider two supervised (mutual information and Fisher's ratio) and two unsupervised (meanmedian and arithmetic mean geometric mean) FS filters. The evaluation was made by assessing the classification accuracy (word recognition error) of three well-known classifiers (knearest neighbors, support vector machines, and dynamic time warping). The key results of this study show that both unsupervised and supervised FS techniques improve on the classification accuracy on both individual and combined modalities. For instance, on the video component, we attain relative performance gains of 36.2% in error rates. FS is also useful as pre-processing for feature fusion. Copyright © 2014 ISCA.

Incremental filter and wrapper approaches for feature discretization

Discrete data representations are necessary, or at least convenient, in many machine learning problems. While feature selection (FS) techniques aim at finding relevant subsets of features, the goal of feature discretization (FD) is to find concise (quantized) data representations, adequate for the learning task at hand. In this paper, we propose two incremental methods for FD. The first method belongs to the filter family, in which the quality of the discretization is assessed by a (supervised or unsupervised) relevance criterion. The second method is a wrapper, where discretized features are assessed using a classifier. Both methods can be coupled with any static (unsupervised or supervised) discretization procedure and can be used to perform FS as pre-processing or post-processing stages. The proposed methods attain efficient representations suitable for binary and multi-class problems with different types of data, being competitive with existing methods. Moreover, using well-known FS methods with the features discretized by our techniques leads to better accuracy than with the features discretized by other methods or with the original features. (C) 2013 Elsevier B.V. All rights reserved.

Viability of the use of thin-film A-SiC:H photodiodes for protein identification

In this paper, we present a multilayer device based on a-Si:H/a-SiC:H that operates as photodetector and optical filter. The use of such device in protein detection applications is relevant in Fluorescence Resonance Energy Transfer (FRET) measurements. This method demands the detection of fluorescent signals located at specific wavelengths bands in the visible part of the electromagnetic spectrum. The device operates in the visible range with a selective sensitivity dependent on electrical and optical bias. Several nanosensors were tested with a commercial spectrophotometer to assess the performance of FRET signals using glucose solutions of different concentrations. The proposed device was used to demonstrate the possibility of FRET signals detection, using visible signals of similar wavelength and intensity. The device sensitivity was tuned to enhance the wavelength band of interest using steady state optical bias at 400 nm. Results show the ability of the device to detect signals in this range. © 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

Insights into the mechanims underlyng the antiproliferative potential of a Co(II) coordination compound bearing 1,10-Phenanthroline-5,6-Dione: DNA and protein interaction studies

The very high antiproliferative activity of [Co(Cl)(H2O)(phendione)(2)][BF4] (phendione is 1,10-phenanthroline-5,6-dione) against three human tumor cell lines (half-maximal inhibitory concentration below 1 mu M) and its slight selectivity for the colorectal tumor cell line compared with healthy human fibroblasts led us to explore the mechanisms of action underlying this promising antitumor potential. As previously shown by our group, this complex induces cell cycle arrest in S phase and subsequent cell death by apoptosis and it also reduces the expression of proteins typically upregulated in tumors. In the present work, we demonstrate that [Co(Cl)(phendione)(2)(H2O)][BF4] (1) does not reduce the viability of nontumorigenic breast epithelial cells by more than 85 % at 1 mu M, (2) promotes the upregulation of proapoptotic Bax and cell-cycle-related p21, and (3) induces release of lactate dehydrogenase, which is partially reversed by ursodeoxycholic acid. DNA interaction studies were performed to uncover the genotoxicity of the complex and demonstrate that even though it displays K (b) (+/- A standard error of the mean) of (3.48 +/- A 0.03) x 10(5) M-1 and is able to produce double-strand breaks in a concentration-dependent manner, it does not exert any clastogenic effect ex vivo, ruling out DNA as a major cellular target for the complex. Steady-state and time-resolved fluorescence spectroscopy studies are indicative of a strong and specific interaction of the complex with human serum albumin, involving one binding site, at a distance of approximately 1.5 nm for the Trp214 indole side chain with log K (b) similar to 4.7, thus suggesting that this complex can be efficiently transported by albumin in the blood plasma.

An unsupervised approach to feature discretization and selection

Many learning problems require handling high dimensional datasets with a relatively small number of instances. Learning algorithms are thus confronted with the curse of dimensionality, and need to address it in order to be effective. Examples of these types of data include the bag-of-words representation in text classification problems and gene expression data for tumor detection/classification. Usually, among the high number of features characterizing the instances, many may be irrelevant (or even detrimental) for the learning tasks. It is thus clear that there is a need for adequate techniques for feature representation, reduction, and selection, to improve both the classification accuracy and the memory requirements. In this paper, we propose combined unsupervised feature discretization and feature selection techniques, suitable for medium and high-dimensional datasets. The experimental results on several standard datasets, with both sparse and dense features, show the efficiency of the proposed techniques as well as improvements over previous related techniques.

Efficient feature selection filters for high-dimensional data

Feature selection is a central problem in machine learning and pattern recognition. On large datasets (in terms of dimension and/or number of instances), using search-based or wrapper techniques can be cornputationally prohibitive. Moreover, many filter methods based on relevance/redundancy assessment also take a prohibitively long time on high-dimensional. datasets. In this paper, we propose efficient unsupervised and supervised feature selection/ranking filters for high-dimensional datasets. These methods use low-complexity relevance and redundancy criteria, applicable to supervised, semi-supervised, and unsupervised learning, being able to act as pre-processors for computationally intensive methods to focus their attention on smaller subsets of promising features. The experimental results, with up to 10(5) features, show the time efficiency of our methods, with lower generalization error than state-of-the-art techniques, while being dramatically simpler and faster.