Average Percentage Of Correct Classification Research Articles

Classification Methods for EEG Patterns of Imaginary Movements

The review focuses on the most promising methods for classifying EEG signals for non-invasive BCIs and theoretical approaches for the successful classification of EEG patterns. The paper provides an overview of articles using Riemannian geometry, deep learning methods and various options for preprocessing and "clustering" EEG signals, for example, common-spatial pattern (CSP). Among other approaches, pre-processing of EEG signals using CSP is often used, both offline and online. The combination of CSP, linear discriminant analysis, support vector machine and neural network (BPNN) made it possible to achieve 91% accuracy for binary classification with exoskeleton control as a feedback. There is very little work on the use of Riemannian geometry online and the best accuracy achieved so far for a binary classification problem is 69.3% in the work. At the same time, in offline testing, the average percentage of correct classification in the considered articles for approaches with CSP – 77.5 ± 5.8%, deep learning networks – 81.7 ± 4.7%, Riemannian geometry – 90.2 ± 6.6%. Due to nonlinear transformations, Riemannian geometry-based approaches and complex deep neural networks provide higher accuracy and better extract of useful information from raw EEG recordings rather than linear CSP transformation. However, in real-time setup, not only accuracy is important, but also a minimum time delay. Therefore, approaches using the CSP transformation and Riemannian geometry with a time delay of less than 500 ms may be in the future advantage.

Monitoring of the phenolic compounds and in vitro antioxidant activity of apple beverages according to geographical origin and their type: A chemometric study

Apple juices (n = 48) were analyzed and classified according to their phenolic composition and in vitro antioxidant activity (DPPH, ABTS, and FRAP), as well as their relationship to their geographical origin and type of juice using a chemometric approach. Total phenolics ranged from 114 to 485 mg CQAE/L and the in vitro antioxidant activity varied from 324 to 9556 μmol TE/L. Significant correlations coefficients (r ≥ 0.32, p < 0.05) were displayed between antioxidant activity and total phenolic compounds and flavonoids. Furthermore, chlorogenic acid and phloridzin also showed correlations with antioxidant activity (r ≥ 0.58 and r = 0.31, respectively). Principal component analysis (PCA) showed an overlap between the samples; and it was not effective enough to distinguish the geographical origin and type of juice. Using hierarchical cluster analysis (HCA) four clusters was suggested, in which cluster 1 showed the highest levels of antioxidant activity (DPPH, ABTS, and FRAP). The SIMCA model was the best of all the developed supervised models at predicting the geographical origin and type of juice, with an average percentage of correct classification that was higher that 80%.

Particle Swarm Optimization of MLP for the identification of factors related to Common Mental Disorders

Social class differences in the prevalence of Common Mental Disorder (CMD) are likely to vary according to time, culture and stage of economic development. The present study aimed to investigate the use of optimization of architecture and weights of Artificial Neural Network (ANN) for identification of the factors related to CMDs. The identification of the factors was possible by optimizing the architecture and weights of the network. The optimization of architecture and weights of ANNs is based on Particle Swarm Optimization with early stopping criteria. This approach achieved a good generalization control, as well as similar or better results than other techniques, but with a lower computational cost, with the ability to generate small networks and with the advantage of the automated architecture selection, which simplify the training process. This paper presents the results obtained in the experiments with ANNs in which it was observed an average percentage of correct classification of individuals with positive diagnostic for the CMDs of 90.59%.

Multi-class classification with probabilistic discriminant partial least squares ( p-DPLS)

This work describes multi-classification based on binary probabilistic discriminant partial least squares ( p-DPLS) models, developed with the strategy one-against-one and the principle of winner-takes-all. The multi-classification problem is split into binary classification problems with p-DPLS models. The results of these models are combined to obtain the final classification result. The classification criterion uses the specific characteristics of an object (position in the multivariate space and prediction uncertainty) to estimate the reliability of the classification, so that the object is assigned to the class with the highest reliability. This new methodology is tested with the well-known Iris data set and a data set of Italian olive oils. When compared with CART and SIMCA, the proposed method has better average performance of classification, besides giving a statistic that evaluates the reliability of classification. For the olive oil set the average percentage of correct classification for the training set was close to 84% with p-DPLS against 75% with CART and 100% with SIMCA, while for the test set the average was close to 94% with p-DPLS as against 50% with CART and 62% with SIMCA.

Geographical classification of Chilean wines by an electronic nose

Geographical classification of Chilean wines by an electronic nose Nicol&aacute;s&nbsp;H Beltr&aacute;n, Manuel A Duarte-Mermoud, Ricardo&nbsp;E Mu&ntilde;ozDepartment of Electrical Engineering, University of Chile, Santiago, ChileAbstract: This paper discusses the classification of Chilean wines by geographical origin based only on aroma information. The varieties of Cabernet Sauvignon, Merlot, and Carm&eacute;n&egrave;re analyzed here are produced in four different valleys in the central part of Chile (Colchagua, Maipo, Maule, and Rapel). Aroma information was obtained with a zNoseTM (fast gas chromatograph) and the data was analyzed by applying wavelet transform for feature extraction followed by an analysis with support vector machines for classification. Two evaluations of the classification technique were performed; the average percentage of correct classification performed on the validation set was obtained by means of cross-validation against the percentage of correct classification obtained on the test set. This developed technique obtained results on classification rates over 94% in both cases. The geographical origin of a Chilean wine can be resolved rapidly with fast gas chromatography and data processing.Keywords: geographical origin, origin denomination, wine classification, pattern recognition, support vector machines, wavelet analysis, feature extraction

AROMA COMPOUNDS AS MARKERS OF WINE QUALITY IN THE CASE OF MALVAZIJA ISTARSKA YOUNG WINE

ABSTRACT In order to investigate the relationship between sensory and gas chromatographic analysis data, and to establish if particular aroma compounds could be used as Malvazija istarska young wine quality markers, 30 samples of Malvazija istarska young wine were subjected to sensory evaluation and gas chromatographic analysis of aroma compounds. Linear regression and principal component analysis were applied in order to establish the relations between sensory evaluation scores and aroma compound concentrations in corresponding wines. The most important contributors to the characterization and differentiation of wines evaluated with higher sensory evaluation scores were isoamyl acetate, ethyl hexanoate, ethyl octanoate, ethyl decanoate, hexanoic acid, octanoic acid and decanoic acid. On the other hand, wines evaluated with lower sensory evaluation scores were characterized and differentiated by higher amounts of isoamyl alcohol and isobutanol. Mentioned esters and acids were also found to be the most important variables for the classification of these wines according to quality using stepwise linear discriminant analysis (SLDA). The evaluation of the SLDA model was performed by cross‐validation, obtaining an average percentage of correct classification of 100.0%, and of correct prediction of 89.3%. The results of this investigation suggest that above‐mentioned esters, acids and alcohols could be used as markers of Malvazija istarska young wine quality. PRACTICAL APPLICATIONSWine aroma is one of the most important parameters responsible for its quality, and hence for consumer acceptance. Sensory evaluation is the most common technique for controlling the organoleptic quality of wine, but it is not always feasible, since it requires specific conditions, and a group of trained expert tasters that can only assess a relatively small number of wine samples for a day. Moreover, the subjectivity of panelists can affect the evaluation in a great measure. The approach proposed in this work suggests the use of a nonsubjective instrumental technique such as gas chromatography for the analysis of volatile aroma compounds as a complementary tool for wine quality control able to predict, confirm or to correct the results of Malvazija istarska young wine sensory evaluation. Although this model is yet far from being practically implemented, information obtained could be used as a starting ground. Moreover, it was assumed that the approach proposed in this work is likely to be applied successfully on the differentiation and classification of other types of wine according to quality.

Phase synchrony and coherence analyses of EEG as tools to discriminate between children with and without attention deficit disorder

Phase synchrony and model-based magnitude squared (MS) coherence analyses of the EEG are investigated for their usefulness in the discrimination between non-ADHD and ADHD children. We observed that phase synchrony is higher – on average – for non-ADHD participants than for the ADHD group. Based on the relatively small sample available to us, we find that a Euclidean distance-based classifier may – with selection of proper electrode pairs, EEG rhythm, and task – discriminate between the EEG of non-ADHD and ADHD children with an average percentage of correct classification of 63.2%. MS coherence was observed to be considerably higher at low frequencies (i.e. below 1 Hz) for ADHD subjects than for the majority of the non-ADHD individuals. We also observed that the θ rhythm MS coherence (approximately 4–6 Hz) is generally higher for non-ADHD subjects than for ADHD individuals. The combination of these effects was observed on four out of the seven non-ADHD subjects and on all six of the ADHD subjects.

Digital image analysis and identification of eggs from bovine parasitic nematodes.

Computer-assisted microscopy and multivariate statistics were used to establish and evaluate a procedure for identification of bovine strongylid eggs. Ostertagia ostertagi, Cooperia oncophora, Haemonchus placei, Trichostrongylus axei, and Oesophagostomum radiatum eggs were obtained from faeces voided by monospecifically infected calves. Images of single eggs (400 x magnification) were recorded by a CCD camera fitted onto a microscope and digitized on a PC. After separation of eggs from the image background, the pixel (picture element) positions of the egg outline were analysed by algorithms to describe size and shape. A stepwise discriminant analysis was subsequently used to select and rank descriptive features of 4207 eggs according to discriminatory power. Classification criteria were developed by linear discrimination analysis on the basis of selected features, and the criteria evaluated by cross-validation. A maximum average percentage of correct classification of 85.8% resulted when nineteen features were employed in a linear classification criterion. The percentages correct classification for each species were: O. ostertagi 76.3%, C. oncophora 90.8%, O. radiatum 87.8%, H. placei 90.1%, and T. axei 83.8%. Classification based on the five most important features gave an overall correct classification of 81.5%. Images of "unknown' eggs could be identified automatically by the classification criteria after procedural steps performed by PC were linked in a batch program.