Adaptive frequency scaled wavelet packet decomposition for frog call classification

Environmental changes have put great pressure on biological systems leading to the rapid decline of biodiversity. To monitor this change and protect biodiversity, animal vocalizations have been widely explored by the aid of deploying acoustic sensors in the field. Consequently, large volumes of acoustic data are collected. However, traditional manual methods that require ecologists to physically visit sites to collect biodiversity data are both costly and time consuming. Therefore it is essential to develop new semi-automated and automated methods to identify species in automated audio recordings. In this study, a novel feature extraction method based on wavelet packet decomposition is proposed for frog call classification. After syllable segmentation, the advertisement call of each frog syllable is represented by a spectral peak track, from which track duration, dominant frequency and oscillation rate are calculated. Then, a k-means clustering algorithm is applied to the dominant frequency, and the centroids of clustering results are used to generate the frequency scale for wavelet packet decomposition (WPD). Next, a new feature set named adaptive frequency scaled wavelet packet decomposition sub-band cepstral coefficients is extracted by performing WPD on the windowed frog calls. Furthermore, the statistics of all feature vectors over each windowed signal are calculated for producing the final feature set. Finally, two well-known classifiers, a k-nearest neighbour classifier and a support vector machine classifier, are used for classification. In our experiments, we use two different datasets from Queensland, Australia (18 frog species from commercial recordings and field recordings of 8 frog species from James Cook University recordings). The weighted classification accuracy with our proposed method is 99.5% and 97.4% for 18 frog species and 8 frog species respectively, which outperforms all other comparable methods.     

Nonlinear wavelet and wavelet packet denoising of electrocardiogram signal

The performance of different wavelet- and wavelet packet-based methods for removing simulated noise was studied using an electrocardiogram (ECG) signal. A non-linear denoising approach was investigated by applying soft and hard thresholding methods, in which thresholds were chosen using four different methods. Coiflet wavelet and wavelet packet functions were used to build up the dyadic wavelet and optimized wavelet packet decompositions. This study involves the quantitative comparison of different denoising approaches by means of optimized error measures and visual inspection of the denoised ECG and the error signal. The localization of the denoising error within the cardiac cycle was studied by visual inspection of the denoised signal and extracting the error measures during the QRS-complex. The results showed that wavelet denoising approaches were generally more efficient than wavelet packet approaches in all cases, but with heuristic sure threshold selection rule as hard thresholding for white noises was used. Denoising errors tend to concentrate within the QRS-area when the wavelet approach was employed. Moreover, soft and hard non-linearities showed different balances in denoising the high-frequency parts of an ECG. Received: 27 April 1998 / Accepted in revised form: 24 November 1998     

Robust PCA and classification in biosciences

MOTIVATION: Principal components analysis (PCA) is a very popular dimension reduction technique that is widely used as a first step in the analysis of high-dimensional microarray data. However, the classical approach that is based on the mean and the sample covariance matrix of the data is very sensitive to outliers. Also, classification methods based on this covariance matrix do not give good results in the presence of outlying measurements. RESULTS: First, we propose a robust PCA (ROBPCA) method for high-dimensional data. It combines projection-pursuit ideas with robust estimation of low-dimensional data. We also propose a diagnostic plot to display and classify the outliers. This ROBPCA method is applied to several bio-chemical datasets. In one example, we also apply a robust discriminant method on the scores obtained with ROBPCA. We show that this combination of robust methods leads to better classifications than classical PCA and quadratic discriminant analysis. AVAILABILITY: All the programs are part of the Matlab Toolbox for Robust Calibration, available at http://www.wis.kuleuven.ac.be/stat/robust.html.     

Integration of wavelet transform with PCA and ANN for metabolomics data-mining

PCA (principal components analysis) and ANN (artificial neural network) are two broadly used pattern recognition methods in metabolomics data-mining. Yet their limitations sometimes are great obstacles for researchers. In this paper the wavelet transform (WT) method was used to integrate with PCA and ANN to improve their performance in manipulating metabolomics data. A dataset was decomposed by wavelets and then reconstructed. The hard thresholding algorithm was used, through which the detail information was discarded, and the entire metabolomics image reconstructed on the significant information. It was supposed that the most relevant information was captured after this process. It was found that, thanks to its ability in denoising data, the WT method could significantly improve the performance of the non-linear essence-extracting method ANN in classifying samples; further integration of WT with PCA showed that WT could greatly enhance the ability of PCA in distinguishing one group of samples from another and also its ability in identifying potential biomarkers. The results highlighted WT as a promising resolution in bridging the gap between huge bytes of data and the instructive biological information.     

Genetic classification of populations using supervised learning

There are many instances in genetics in which we wish to determine whether two candidate populations are distinguishable on the basis of their genetic structure. Examples include populations which are geographically separated, case-control studies and quality control (when participants in a study have been genotyped at different laboratories). This latter application is of particular importance in the era of large scale genome wide association studies, when collections of individuals genotyped at different locations are being merged to provide increased power. The traditional method for detecting structure within a population is some form of exploratory technique such as principal components analysis. Such methods, which do not utilise our prior knowledge of the membership of the candidate populations. are termed unsupervised. Supervised methods, on the other hand are able to utilise this prior knowledge when it is available.In this paper we demonstrate that in such cases modern supervised approaches are a more appropriate tool for detecting genetic differences between populations. We apply two such methods, (neural networks and support vector machines) to the classification of three populations (two from Scotland and one from Bulgaria). The sensitivity exhibited by both these methods is considerably higher than that attained by principal components analysis and in fact comfortably exceeds a recently conjectured theoretical limit on the sensitivity of unsupervised methods. In particular, our methods can distinguish between the two Scottish populations, where principal components analysis cannot. We suggest, on the basis of our results that a supervised learning approach should be the method of choice when classifying individuals into pre-defined populations, particularly in quality control for large scale genome wide association studies.     

Estimating and minimizing movement artifacts in surface electromyogram

While recording surface electromyography [sEMG], it is possible to record the electrical activities coming from the muscles and transients in the half-cell potential at the electrode–electrolyte interface due to micromovements of the electrode–skin interface. Separating the two sources of electrical activity usually fails due to the overlapping frequency characteristics of the signals. This paper aims to develop a method that detects movement artifacts and suggests a minimization technique. Towards that aim, we first estimated the frequency characteristics of movement artifacts under various static and dynamic experimental conditions. We found that the extent of the movement artifact depended on the nature of the movement and varied from person to person. Our study's highest movement artifact frequency for the stand position was 10 Hz, tiptoe 22, walk 32, run 23, jump from box 41, and jump up and down 40 Hz. Secondly, using a 40 Hz highpass filter, we cut out most of the frequencies belonging to the movement artifacts. Finally, we checked whether the latencies and amplitudes of reflex and direct muscle responses were still observed in the highpass-filtered sEMG. We showed that the 40 Hz highpass filter did not significantly alter reflex and direct muscle variables. Therefore, we recommend that researchers who use sEMG under similar conditions employ the recommended level of highpass filtering to reduce movement artifacts from their records. However, suppose different movement conditions are used. In that case, it is best to estimate the frequency characteristics of the movement artifact before applying any highpass filtering to minimize movement artifacts and their harmonics from sEMG.     

Multi-channel surface EMG classification using support vector machines and signal-based wavelet optimization

The study proposes a method for supervised classification of multi-channel surface electromyographic signals with the aim of controlling myoelectric prostheses. The representation space is based on the discrete wavelet transform (DWT) of each recorded EMG signal using unconstrained parameterization of the mother wavelet. The classification is performed with a support vector machine (SVM) approach in a multi-channel representation space. The mother wavelet is optimized with the criterion of minimum classification error, as estimated from the learning signal set. The method was applied to the classification of six hand movements with recording of the surface EMG from eight locations over the forearm. Misclassification rate in six subjects using the eight channels was (mean ± S.D.) 4.7 ± 3.7% with the proposed approach while it was 11.1 ± 10.0% without wavelet optimization (Daubechies wavelet). The DWT and SVM can be implemented with fast algorithms, thus, the method is suitable for real-time implementation.     

Towards a supervised classification of neocortical interneuron morphologies

Background

The challenge of classifying cortical interneurons is yet to be solved. Data-driven classification into established morphological types may provide insight and practical value.

Results

We trained models using 217 high-quality morphologies of rat somatosensory neocortex interneurons reconstructed by a single laboratory and pre-classified into eight types. We quantified 103 axonal and dendritic morphometrics, including novel ones that capture features such as arbor orientation, extent in layer one, and dendritic polarity. We trained a one-versus-rest classifier for each type, combining well-known supervised classification algorithms with feature selection and over- and under-sampling. We accurately classified the nest basket, Martinotti, and basket cell types with the Martinotti model outperforming 39 out of 42 leading neuroscientists. We had moderate accuracy for the double bouquet, small and large basket types, and limited accuracy for the chandelier and bitufted types. We characterized the types with interpretable models or with up to ten morphometrics.

Conclusion

Except for large basket, 50 high-quality reconstructions sufficed to learn an accurate model of a type. Improving these models may require quantifying complex arborization patterns and finding correlates of bouton-related features. Our study brings attention to practical aspects important for neuron classification and is readily reproducible, with all code and data available online.

     

Reproducibility of the mean power frequency of the surface electromyogram

There is evidence in the literature that the decrease of mean power frequency (MPF) during exercise is greater as a muscle become more fatigued. After strenuous exercise this phenomenon can last several days. It is usually assumed, however, that the MPF has a good reproducibility. In this study the reproducibility of the MPF of the surface electromyogram of the biceps brachii muscle was investigated for five subjects on 5 successive days. Force level, muscle length and skin temperature were kept constant. The results show that interindividual differences in MPF were large (SD 11.5 Hz). However, during these 5 days, the range in MPF for individual subjects was small. The SD of the trials within subjects and days was 2.0 Hz, while the SD trials excluded). It is hypothesized that this SD may be due to variations in the electrode replacement. It is concluded that the variability in MPF for a subject is small compared to the decrease of the MPF associated with muscle fatigue and which can therefore be determined reliably during longitudinal studies.     

Influence of amplitude cancellation on the simulated surface electromyogram.

The purpose of the study was to quantify the influence of selected motor unit properties and patterns of activity on amplitude cancellation in the simulated surface electromyogram (EMG). The study involved computer simulations of a motor unit population with physiologically defined recruitment and rate coding characteristics that activated muscle fibers whose potentials were recorded on the skin over the muscle. Amplitude cancellation was quantified as the percent difference in signal amplitude when motor unit potentials were summed before and after rectification. The simulations involved varying the level of activation for the motor unit population, the recording configuration, the upper limit of motor unit recruitment, peak discharge rates, the amount of motor unit synchronization, muscle fiber length, the thickness of the subcutaneous tissue, and the motor unit properties that change with advancing age. The results confirmed a previous experimental report (Day SJ and Hulliger M, J Neurophysiol 86: 2144-2158, 2001) that amplitude cancellation in the surface EMG can reach 62% at maximal activation. A decrease in the range of amplitudes of the motor unit potentials, as can occur during fatiguing contractions, increased amplitude cancellation up to approximately 85%. Differences in the amount of amplitude cancellation were observed across all simulated conditions, and resulted in substantial changes in the absolute magnitude of the EMG signal. The most profound factors influencing amplitude cancellation were the number of active motor units and the duration of the action potentials. The effects of amplitude cancellation were minimal (<5%) when the EMG amplitude was normalized to maximal values, with the exception of variations in peak discharge rate and recruitment range, which resulted in differences up to 17% in the normalized EMG signal across conditions. These results indicate the amount of amplitude cancellation that can occur in various experimental conditions and its influence on absolute and relative measures of EMG amplitude.     

Changes in the surface electromyogram during increasing isometric shoulder forward flexions

When using electromyographic techniques in the evaluation of muscular load it is necessary to determine the mathematical relationship between the torque and the amplitude of the electromyographic signal. Isometric gradually increasing contractions up to 100% MVC can then be used. Often more than linear increases for the amplitude (RMS)--force regression have been reported. The present study was designed to test whether changes in power spectral density function take place during a gradually increasing isometric contraction (duration 10 s). Twenty-two clinically healthy females performed an increasing isometric shoulder forward flexion for 10 s using an isokinetic dynamometer. Electromyographic activity was measured in trapezius, deltoid, infraspinatus and biceps brachii using surface electrodes. Mean torque values were determined together with mean power frequency (MPF) and root mean square values (RMS) from the EMG signals for each 256 ms period. The RMS-torque regressions showed higher regression coefficients during the 6th to 9th sec than during the first 5 s. No significant correlation existed between MPF for the four muscles and the torque. A gradual decrease in MPF was generally found from the 6th s. It is concluded that this decrease in power spectral density function might have contributed to the significantly higher regression coefficient for the RMS torque regression at the high output part of the gradually increasing isometric contraction.     

Improvement of image classification using wavelet coefficients with structured-based neural network

Image classification is a challenging problem in organizing a large image database. However, an effective method for such an objective is still under investigation. A method based on wavelet analysis to extract features for image classification is presented in this paper. After an image is decomposed by wavelet, the statistics of its features can be obtained by the distribution of histograms of wavelet coefficients, which are respectively projected onto two orthogonal axes, i.e., x and y directions. Therefore, the nodes of tree representation of images can be represented by the distribution. The high level features are described in low dimensional space including 16 attributes so that the computational complexity is significantly decreased. 2,800 images derived from seven categories are used in experiments. Half of the images were used for training neural network and the other images used for testing. The features extracted by wavelet analysis and the conventional features are used in the experiments to prove the efficacy of the proposed method. The classification rate on the training data set with wavelet analysis is up to 91%, and the classification rate on the testing data set reaches 89%. Experimental results show that our proposed approach for image classification is more effective.     

A comparison of supervised learning techniques in the classification of bat echolocation calls

Today's acoustic monitoring devices are capable of recording and storing tremendous amounts of data. Until recently, the classification of animal vocalizations from field recordings has been relegated to qualitative approaches. For large-scale acoustic monitoring studies, qualitative approaches are very time-consuming and suffer from the bias of subjectivity. Recent developments in supervised learning techniques can provide rapid, accurate, species-level classification of bioacoustics data. We compared the classification performances of four supervised learning techniques (random forests, support vector machines, artificial neural networks, and discriminant function analysis) for five different classification tasks using bat echolocation calls recorded by a popular frequency-division bat detector. We found that all classifiers performed similarly in terms of overall accuracy with the exception of discriminant function analysis, which had the lowest average performance metrics. Random forests had the advantage of high sensitivities, specificities, and predictive powers across the majority of classification tasks, and also provided metrics for determining the relative importance of call features in distinguishing between groups. Overall classification accuracy for each task was slightly lower than reported accuracies using calls recorded by time-expansion detectors. Myotis spp. were particularly difficult to separate; classifiers performed best when members of this genus were combined in genus-level classification and analyzed separately at the level of species. Additionally, we identified and ranked the relative contributions of all predictor features to classifier accuracy and found measurements of frequency, total call duration, and characteristic slope to be the most important contributors to classification success. We provide recommendations to maximize accuracy and efficiency when analyzing acoustic data, and suggest an application of automated bioacoustics monitoring to contribute to wildlife monitoring efforts.     

PCP: a program for supervised classification of gene expression profiles

PCP (Pattern Classification Program) is an open-source machine learning program for supervised classification of patterns (vectors of measurements). The principal use of PCP in bioinformatics is design and evaluation of classifiers for use in clinical diagnostic tests based on measurements of gene expression. PCP implements leading pattern classification and gene selection algorithms and incorporates cross-validation estimation of classifier performance. Importantly, the implementation integrates gene selection and class prediction stages, which is vital for computing reliable performance estimates in small-sample scenarios. Additionally, the program includes automated and efficient model selection (optimization of parameters) for support vector machine (SVM) classifier. The distribution includes Linux and Windows/Cygwin binaries. The program can easily be ported to other platforms. AVAILABILITY: Free download at http://pcp.sourceforge.net     

Manifestation of morphofunctional specificity of muscles in spectral evaluations of the surface electromyogram

It is shown that, when the global recording method is used, the muscles of upper limbs (in comparison with lower ones) and of distal parts of limbs (in comparison with proximal ones) generate wider surface electromyograms (EMGs) with a higher dominant frequency and lower saturation of power spectra owing to the specific features of the morphofunctional organization of both the muscles proper and corticospinal systems controlling their voluntary contractive activity. There is an interrelation among different spectral and amplitude-frequency response (EMG analysis according to Willison) characteristics of the surface EMG. Limb lengthening may be considered, on the basis of criteria of changes in spectral characteristics of the EMG of lengthened segment muscles and results of its analysis according to Willison, as a model of a decrease in reliability of corticospinal connections.Translated from Fiziologiya Cheloveka, Vol. 31, No. 1, 2005, pp. 66–76.Original Russian Text Copyright © 2005 by Shein, Krivoruchko, Saifutdinov.     

Spherical classification of wavelet transformed EMG intensity patterns

Electromyograms of different muscles can be submitted to a wavelet-transform and arranged in a Multi-Muscle Pattern (MMP). The MMP represents the intensity of the EMG signals of a number of muscles simultaneously in time/frequency space. As previously shown, the MMPs can be represented by points in an Euclidian vector space that was called pattern space. The variability of the MMPs is represented by the distribution of the scattered points in pattern space. The purpose of this study was to investigate the distribution of the points and use the properties of the distribution to classify MMPs. The first task was to test whether the points representing a group of MMPs were located between the inner and outer boundary of a sphere-like domain in whitened pattern space as theoretically predicted. The mean of these points and thus of the MMPs is represented by a point at the center of the sphere. The hypothesis was that the spheres representing points of the MMPs of barefoot and shod runners were sufficiently separated and distinguishable in pattern space to allow classification of the runners according to their shod condition. The results confirmed the hypothesis and revealed that the recognition rate was over 80%. One can conclude and generalize that the points representing MMPs recorded for a certain condition reside between the inner and outer boundary of the sphere. The classification based on the spherical feature represents a much better discrimination than one based on the distance from the mean.     

Classification of epileptic seizures using wavelet packet log energy and norm entropies with recurrent Elman neural network classifier

Electroencephalogram shortly termed as EEG is considered as the fundamental segment for the assessment of the neural activities in the brain. In cognitive neuroscience domain, EEG-based assessment method is found to be superior due to its non-invasive ability to detect deep brain structure while exhibiting superior spatial resolutions. Especially for studying the neurodynamic behavior of epileptic seizures, EEG recordings reflect the neuronal activity of the brain and thus provide required clinical diagnostic information for the neurologist. This specific proposed study makes use of wavelet packet based log and norm entropies with a recurrent Elman neural network (REN) for the automated detection of epileptic seizures. Three conditions, normal, pre-ictal and epileptic EEG recordings were considered for the proposed study. An adaptive Weiner filter was initially applied to remove the power line noise of 50 Hz from raw EEG recordings. Raw EEGs were segmented into 1 s patterns to ensure stationarity of the signal. Then wavelet packet using Haar wavelet with a five level decomposition was introduced and two entropies, log and norm were estimated and were applied to REN classifier to perform binary classification. The non-linear Wilcoxon statistical test was applied to observe the variation in the features under these conditions. The effect of log energy entropy (without wavelets) was also studied. It was found from the simulation results that the wavelet packet log entropy with REN classifier yielded a classification accuracy of 99.70 % for normal-pre-ictal, 99.70 % for normal-epileptic and 99.85 % for pre-ictal-epileptic.     

Use of DNA reassociation in bacterial classification

The reassociation properties of DNA provide invaluable taxonomic tools. Different methods may give different reassociation values. However, the thermal stability of reassociated DNA strands (a measurement that seems independent of method) is useful in delineating genomic species. Although many phenotypically defined species have been confirmed by DNA reassociation, some medically important genomic species previously had been split into several nomenspecies on the basis of a few characteristics whereas some environmental genomic species had been lumped into unidentifiable aggregates. It might take some time before the nomenclature can be adapted to new taxonomic findings.