Method for exploratory cluster analysis and visualisation of single-trial ERP ensembles

Background: The validity of ensemble averaging on event-related potential (ERP) data has been questioned, due to its assumption that the ERP is identical across trials. Thus, there is a need for preliminary testing for cluster structure in the data. New method: We propose a complete pipeline for the cluster analysis of ERP data. To increase the signalto-noise (SNR) ratio of the raw single-trials, we used a denoising method based on Empirical Mode Decomposition (EMD). Next, we used a bootstrap-based method to determine the number of clusters, through a measure called the Stability Index (SI). We then used a clustering algorithm based on a Genetic Algorithm (GA)to define initial cluster centroids for subsequent k-means clustering. Finally, we visualised the clustering results through a scheme based on Principal Component Analysis (PCA). Results: After validating the pipeline on simulated data, we tested it on data from two experiments – a P300 speller paradigm on a single subject and a language processing study on 25 subjects. Results revealed evidence for the existence of 6 clusters in one experimental condition from the language processing study. Further, a two-way chi-square test revealed an influence of subject on cluster membership.

Visual Object Recognition Using Deformable Models of Vehicles

This workshop paper reports recent developments to a vision system for traffic interpretation which relies extensively on the use of geometrical and scene context. Firstly, a new approach to pose refinement is reported, based on forces derived from prominent image derivatives found close to an initial hypothesis. Secondly, a parameterised vehicle model is reported, able to represent different vehicle classes. This general vehicle model has been fitted to sample data, and subjected to a Principal Component Analysis to create a deformable model of common car types having 6 parameters. We show that the new pose recovery technique is also able to operate on the PCA model, to allow the structure of an initial vehicle hypothesis to be adapted to fit the prevailing context. We report initial experiments with the model, which demonstrate significant improvements to pose recovery.

Arctic oscillation or North Atlantic oscillation?

The definition and interpretation of the Arctic oscillation (AO) are examined and compared with those of the North Atlantic oscillation (NAO). It is shown that the NAO reflects the correlations between the surface pressure variability at its centers of action, whereas this is not the case for the AO. The NAO pattern can be identified in a physically consistent way in principal component analysis applied to various fields in the Euro-Atlantic region. A similar identification is found in the Pacific region for the Pacific–North American (PNA) pattern, but no such identification is found here for the AO. The AO does reflect the tendency for the zonal winds at 35° and 55°N to anticorrelate in both the Atlantic and Pacific regions associated with the NAO and PNA. Because climatological features in the two ocean basins are at different latitudes, the zonally symmetric nature of the AO does not mean that it represents a simple modulation of the circumpolar flow. An increase in the AO or NAO implies strong, separated tropospheric jets in the Atlantic but a weakened Pacific jet. The PNA has strong related variability in the Pacific jet exit, but elsewhere the zonal wind is similar to that related to the NAO. The NAO-related zonal winds link strongly through to the stratosphere in the Atlantic sector. The PNA-related winds do so in the Pacific, but to a lesser extent. The results suggest that the NAO paradigm may be more physically relevant and robust for Northern Hemisphere variability than is the AO paradigm. However, this does not disqualify many of the physical mechanisms associated with annular modes for explaining the existence of the NAO.

Some factors affecting variation in milk yield in crossbred dairy cows on smallholder farms in north-east Tanzania

A 2-year longitudinal survey was carried out to investigate factors affecting milk yield in crossbred cows on smallholder farms in and around an urban centre. Sixty farms were visited at approximately 2-week intervals and details of milk yield, body condition score (BCS) and heart girth measurements were collected. Fifteen farms were within the town (U), 23 farms were approximately 5 km from town (SU), and 22 farms approximately 10 km from town (PU). Sources of variation in milk yield were investigated using a general linear model by a stepwise forward selection and backward elimination approach to judge important independent variables. Factors considered for the first step of formulation of the model included location (PU, SU and U), calving season, BCS at calving, at 3 months postpartum and at 6 months postpartum, calving year, herd size category, source of labour (hired and family labour), calf rearing method (bucket and partial suckling) and parity number of the cow. Daily milk yield (including milk sucked by calves) was determined by calving year (p < 0.0001), calf rearing method (p = 0.044) and BCS at calving (p < 0.0001). Only BCS at calving contributed to variation in volume of milk sucked by the calf, lactation length and lactation milk yield. BCS at 3 months after calving was improved on farms where labour was hired (p = 0.041) and BCS change from calving to 6 months was more than twice as likely to be negative on U than SU and PU farms. It was concluded that milk production was predominantly associated with BCS at calving, lactation milk yield increasing quadratically from score 1 to 3. BCS at calving may provide a simple, single indicator of the nutritional status of a cow population.

Some factors affecting reproductive success in crossbred dairy cows on smallholder farms in coastal north-east Tanzania

A 2-year longitudinal survey was carried out to investigate factors affecting reproduction in crossbred cows on smallholder farms in and around an urban centre. Sixty farms were visited at approximately 2-week intervals and details of reproductive traits and body condition score (BCS) were collected. Fifteen farms were within the town (U), 23 farms were approximately 5 km from town (SU), and 22 farms approximately 10 km from town (PU). Sources of variation in reproductive traits were investigated using a general linear model (GLM) by a stepwise forward selection and backward elimination approach to judge important independent variables. Factors considered for the first step of formulation of the model included location (PU, SU and U), type of insemination, calving season, BCS at calving, at 3 months postpartum and at 6 months postpartum, calving year, herd size category, source of labour (hired and family labour), calf rearing method (bucket and partial suckling) and parity number of the cow. The effects of the independent variables identified were then investigated using a non-parametric survival technique. The number of days to first oestrus was increased on the U site (p = 0.045) and when family labour was used (p = 0.02). The non-parametric test confirmed the effect of site (p = 0.059), but effect of labour was not significant. The number of days from calving to conception was reduced by hiring labour (p = 0.003) and using natural service (p = 0.028). The non-parametric test confirmed the effects of type of insemination (p = 0.0001) while also identifying extended calving intervals on U and SU sites (p = 0.014). Labour source was again non-significant. Calving interval was prolonged on U and SU sites (p = 0.021), by the use of AI (p = 0.031) and by the use of family labour (p = 0.001). The non-parametric test confirmed the effect of site (p = 0.008) and insemination type (p > 0.0001) but not of labour source. It was concluded that under favourable conditions (PU site, hired labour and natural service) calving intervals of around 440 days could be achieved.

Chemical characterisation of wild populations of Thymus from different climatic regions in southeast Spain

Thymus is taxonomically a very complex genus with a high frequency of hybridisation and introgression among sympatric species. The variation in accumulation of leaf-surface flavonoids was investigated in 71 wild populations of Thymus front different putative hybrid swarm areas in Andalucia, Spain. Twenty-two flavones, five flavanones, two dihydroflavonols, a flavonol and two unknowns were detected by HPLC-DAD combined with LC-APCI-MS analysis. The majority of compounds were flavones with a lutelin-type substitution of the B-ring, in contrast to previous reports on Macedonian taxa, which predominantly accumulate flavones with apigenin-type substitution of the B-ring. Anatomical and morphometric studies, supported by cluster analysis, identified pure Thymus hyemalis and Thymus baeticus populations, and a large number of putative hybrids. Flavonoid variation was closely related to morphological variation in all populations and is suspected to be a result of genetic polymorphism. Principal component analysis identified the presence of species-specific and geographically linked chemotypes and putative hybrids with mixed morphological and chemical characteristics. Qualitative and quantitative flavonoid accumulation appears to be genetically regulated, while external factors play a secondary role. Flavonoid profiles can thus provide diagnostic markers for the taxonomy of Thymus and are also useful in detecting hybridising taxa. (C) 2007 Elsevier Ltd. All rights reserved.

Comparison of small and large deformation rheological properties of wheat dough and gluten

The rheological properties of dough and gluten are important for end-use quality of flour but there is a lack of knowledge of the relationships between fundamental and empirical tests and how they relate to flour composition and gluten quality. Dough and gluten from six breadmaking wheat qualities were subjected to a range of rheological tests. Fundamental (small-deformation) rheological characterizations (dynamic oscillatory shear and creep recovery) were performed on gluten to avoid the nonlinear influence of the starch component, whereas large deformation tests were conducted on both dough and gluten. A number of variables from the various curves were considered and subjected to a principal component analysis (PCA) to get an overview of relationships between the various variables. The first component represented variability in protein quality, associated with elasticity and tenacity in large deformation (large positive loadings for resistance to extension and initial slope of dough and gluten extension curves recorded by the SMS/Kieffer dough and gluten extensibility rig, and the tenacity and strain hardening index of dough measured by the Dobraszczyk/Roberts dough inflation system), the elastic character of the hydrated gluten proteins (large positive loading for elastic modulus [G'], large negative loadings for tan delta and steady state compliance [J(e)(0)]), the presence of high molecular weight glutenin subunits (HMW-GS) 5+10 vs. 2+12, and a size distribution of glutenin polymers shifted toward the high-end range. The second principal component was associated with flour protein content. Certain rheological data were influenced by protein content in addition to protein quality (area under dough extension curves and dough inflation curves [W]). The approach made it possible to bridge the gap between fundamental rheological properties, empirical measurements of physical properties, protein composition, and size distribution. The interpretation of this study gave indications of the molecular basis for differences in breadmaking performance.

Colonic microbiota signatures across five northern European countries

The composition of the colonic microbiota of 91 northern Europeans was characterized by fluorescent in situ hybridization using 18 phylogenetic probes. On average 75% of the bacteria were identified, and large interindividual variations were observed. Clostridium coccoides and Clostridium leptum were the dominant groups (28.0% and 25.2%), followed by the Bacteroides (8.5%). According to principal component analysis, no significant grouping with respect to geographic origin, age, or gender was observed.

Sparse kernel modelling: a unified approach

A unified approach is proposed for sparse kernel data modelling that includes regression and classification as well as probability density function estimation. The orthogonal-least-squares forward selection method based on the leave-one-out test criteria is presented within this unified data-modelling framework to construct sparse kernel models that generalise well. Examples from regression, classification and density estimation applications are used to illustrate the effectiveness of this generic sparse kernel data modelling approach.

Sparse kernel density estimator using orthogonal regression based on D-optimality experimental design

A novel sparse kernel density estimator is derived based on a regression approach, which selects a very small subset of significant kernels by means of the D-optimality experimental design criterion using an orthogonal forward selection procedure. The weights of the resulting sparse kernel model are calculated using the multiplicative nonnegative quadratic programming algorithm. The proposed method is computationally attractive, in comparison with many existing kernel density estimation algorithms. Our numerical results also show that the proposed method compares favourably with other existing methods, in terms of both test accuracy and model sparsity, for constructing kernel density estimates.

A semi-blind reception scheme based on ICA for multiuser MIMO-STBC system

When the orthogonal space-time block code (STBC), or the Alamouti code, is applied on a multiple-input multiple-output (MIMO) communications system, the optimum reception can be achieved by a simple signal decoupling at the receiver. The performance, however, deteriorates significantly in presence of co-channel interference (CCI) from other users. In this paper, such CCI problem is overcome by applying the independent component analysis (ICA), a blind source separation algorithm. This is based on the fact that, if the transmission data from every transmit antenna are mutually independent, they can be effectively separated at the receiver with the principle of the blind source separation. Then equivalently, the CCI is suppressed. Although they are not required by the ICA algorithm itself, a small number of training data are necessary to eliminate the phase and order ambiguities at the ICA outputs, leading to a semi-blind approach. Numerical simulation is also shown to verify the proposed ICA approach in the multiuser MIMO system.

Generative topographic mapping applied to clustering and visualization of motor unit action potentials

The identification and visualization of clusters formed by motor unit action potentials (MUAPs) is an essential step in investigations seeking to explain the control of the neuromuscular system. This work introduces the generative topographic mapping (GTM), a novel machine learning tool, for clustering of MUAPs, and also it extends the GTM technique to provide a way of visualizing MUAPs. The performance of GTM was compared to that of three other clustering methods: the self-organizing map (SOM), a Gaussian mixture model (GMM), and the neural-gas network (NGN). The results, based on the study of experimental MUAPs, showed that the rate of success of both GTM and SOM outperformed that of GMM and NGN, and also that GTM may in practice be used as a principled alternative to the SOM in the study of MUAPs. A visualization tool, which we called GTM grid, was devised for visualization of MUAPs lying in a high-dimensional space. The visualization provided by the GTM grid was compared to that obtained from principal component analysis (PCA). (c) 2005 Elsevier Ireland Ltd. All rights reserved.

Fully complex-valued radial basis function networks: orthogonal least squares regression and classification

We consider a fully complex-valued radial basis function (RBF) network for regression and classification applications. For regression problems, the locally regularised orthogonal least squares (LROLS) algorithm aided with the D-optimality experimental design, originally derived for constructing parsimonious real-valued RBF models, is extended to the fully complex-valued RBF (CVRBF) network. Like its real-valued counterpart, the proposed algorithm aims to achieve maximised model robustness and sparsity by combining two effective and complementary approaches. The LROLS algorithm alone is capable of producing a very parsimonious model with excellent generalisation performance while the D-optimality design criterion further enhances the model efficiency and robustness. By specifying an appropriate weighting for the D-optimality cost in the combined model selecting criterion, the entire model construction procedure becomes automatic. An example of identifying a complex-valued nonlinear channel is used to illustrate the regression application of the proposed fully CVRBF network. The proposed fully CVRBF network is also applied to four-class classification problems that are typically encountered in communication systems. A complex-valued orthogonal forward selection algorithm based on the multi-class Fisher ratio of class separability measure is derived for constructing sparse CVRBF classifiers that generalise well. The effectiveness of the proposed algorithm is demonstrated using the example of nonlinear beamforming for multiple-antenna aided communication systems that employ complex-valued quadrature phase shift keying modulation scheme. (C) 2007 Elsevier B.V. All rights reserved.

A kernel-based two-class classifier for imbalanced data sets

Many kernel classifier construction algorithms adopt classification accuracy as performance metrics in model evaluation. Moreover, equal weighting is often applied to each data sample in parameter estimation. These modeling practices often become problematic if the data sets are imbalanced. We present a kernel classifier construction algorithm using orthogonal forward selection (OFS) in order to optimize the model generalization for imbalanced two-class data sets. This kernel classifier identification algorithm is based on a new regularized orthogonal weighted least squares (ROWLS) estimator and the model selection criterion of maximal leave-one-out area under curve (LOO-AUC) of the receiver operating characteristics (ROCs). It is shown that, owing to the orthogonalization procedure, the LOO-AUC can be calculated via an analytic formula based on the new regularized orthogonal weighted least squares parameter estimator, without actually splitting the estimation data set. The proposed algorithm can achieve minimal computational expense via a set of forward recursive updating formula in searching model terms with maximal incremental LOO-AUC value. Numerical examples are used to demonstrate the efficacy of the algorithm.

Prediction of Parkinson’s Disease tremor onset using a radial basis function neural network based on particle swarm optimization

Deep Brain Stimulation (DBS) has been successfully used throughout the world for the treatment of Parkinson's disease symptoms. To control abnormal spontaneous electrical activity in target brain areas DBS utilizes a continuous stimulation signal. This continuous power draw means that its implanted battery power source needs to be replaced every 18–24 months. To prolong the life span of the battery, a technique to accurately recognize and predict the onset of the Parkinson's disease tremors in human subjects and thus implement an on-demand stimulator is discussed here. The approach is to use a radial basis function neural network (RBFNN) based on particle swarm optimization (PSO) and principal component analysis (PCA) with Local Field Potential (LFP) data recorded via the stimulation electrodes to predict activity related to tremor onset. To test this approach, LFPs from the subthalamic nucleus (STN) obtained through deep brain electrodes implanted in a Parkinson patient are used to train the network. To validate the network's performance, electromyographic (EMG) signals from the patient's forearm are recorded in parallel with the LFPs to accurately determine occurrences of tremor, and these are compared to the performance of the network. It has been found that detection accuracies of up to 89% are possible. Performance comparisons have also been made between a conventional RBFNN and an RBFNN based on PSO which show a marginal decrease in performance but with notable reduction in computational overhead.