Average Classification Accuracy Rate Research Articles

The Construction of a Majority-Voting Ensemble Based on the Interrelation and Amount of Information of Features

AbstractIn this paper, we introduced a new ensemble learning algorithm called VIBES, which is better in terms of performance when compared to 85 machine learning algorithms in WEKA tool. This new algorithm is based on three major processes: (i) making an assumption regarding whether features are dependent on or independent of each other, (ii) computing the amount of information of features when it is assumed that they are dependent on each other and then sorting them in a descending manner based on the amount of information, (iii) speeding up the algorithm by optimizing the forward search algorithm that is used in the construction of the final hypothesis from base learner hypotheses. As a result of these processes, it has been seen in the experiments that choosing the relevant assumption can boost learning performance if features are independent of each other; considering features according to the amount of information provides high accuracy and diversity of base learner models. According to experiment results, the algorithm that has been developed has the highest average classification accuracy rate across the 33 datasets. The highest and the lowest average classification accuracy rates have been found to be 89.80 and 78.03%, respectively.

Research on fault analysis of pumping station units based on stack auto-encoder

The stack Auto-encoder is applied to the fault analysis of pumping station units, and a fault analysis model based on stack automatic encoder is constructed. The model consists of an input layer, 3 hidden layers and an output layer to realize feature extraction and dimension reduction of the monitoring data of the pumping station unit. The network uses unsupervised layer-by-layer greedy methods to train, then uses backpropagation algorithms to optimize network parameters and uses the softmax classifier for classification. The experiment proves that the average classification accuracy rate of unit failure and different working conditions reaches 79.8890%, which can provide a certain reference for unit failure analysis.

A Multi-Feature Fusion Based on Transfer Learning for Chicken Embryo Eggs Classification

The fertility detection of Specific Pathogen Free (SPF) chicken embryo eggs in vaccine preparation is a challenging task due to the high similarity among six kinds of hatching embryos (weak, hemolytic, crack, infected, infertile, and fertile). This paper firstly analyzes two classification difficulties of feature similarity with subtle variations on six kinds of five- to seven-day embryos, and proposes a novel multi-feature fusion based on Deep Convolutional Neural Network (DCNN) architecture in a small dataset. To avoid overfitting, data augmentation is employed to generate enough training images after the Region of Interest (ROI) of original images are cropped. Then, all the augmented ROI images are fed into pretrained AlexNet and GoogLeNet to learn the discriminative deep features by transfer learning, respectively. After the local features of Speeded Up Robust Feature (SURF) and Histogram of Oriented Gradient (HOG) are extracted, the multi-feature fusion with deep features and local features is implemented. Finally, the Support Vector Machine (SVM) is trained with the fused features. The verified experiments show that this proposed method achieves an average classification accuracy rate of 98.4%, and that the proposed transfer learning has superior generalization and better classification performance for small-scale agricultural image samples.

Classification of surface electromyogram signals based on directed acyclic graphs and support vector machines

This paper presents a novel classification approach for surface electromyogram (sEMG) signals. The proposed classification approach involves two steps: (1) feature extraction from an sEMG, in which a 7-dimensional feature vector is extracted from 27 types of features of the sEMG by linear discriminant analysis (LDA), and (2) a novel classifier, DAGSVMerr, based on a directed acyclic graph (DAG) and support vector machine (SVM), in which a separability measure function based on erroneous recognition rates (ERRs) is defined to determine the initial operation list. The proposed approach takes advantage of the feedback idea to improve the performance of the classification. The experimental results show that the proposed approach has a better performance than traditional methods, and it achieves an average classification accuracy rate of 99.4%$\, {\pm \, }$1.3% with an error rate of 0.6%. Correct classification rates of the proposed approach are very high, and the approach can be utilized to recognize gesture instructions by analyzing sEMG signals in gesture equipment control studies.

Olfactory Recognition Based on EEG Gamma-Band Activity.

There are various kinds of brain monitoring techniques, including local field potential, near-infrared spectroscopy, magnetic resonance imaging (MRI), positron emission tomography, functional MRI, electroencephalography (EEG), and magnetoencephalography. Among those techniques, EEG is the most widely used one due to its portability, low setup cost, and noninvasiveness. Apart from other advantages, EEG signals also help to evaluate the ability of the smelling organ. In such studies, EEG signals, which are recorded during smelling, are analyzed to determine the subject lacks any smelling ability or to measure the response of the brain. The main idea of this study is to show the emotional difference in EEG signals during perception of valerian, lotus flower, cheese, and rosewater odors by the EEG gamma wave. The proposed method was applied to the EEG signals, which were taken from five healthy subjects in the conditions of eyes open and eyes closed at the Swiss Federal Institute of Technology. In order to represent the signals, we extracted features from the gamma band of the EEG trials by continuous wavelet transform with the selection of Morlet as a wavelet function. Then the [Formula: see text]-nearest neighbor algorithm was implemented as the classifier for recognizing the EEG trials as valerian, lotus flower, cheese, and rosewater. We achieved an average classification accuracy rate of 87.50% with the 4.3 standard deviation value for the subjects in eyes-open condition and an average classification accuracy rate of 94.12% with the 2.9 standard deviation value for the subjects in eyes-closed condition. The results prove that the proposed continuous wavelet transform-based feature extraction method has great potential to classify the EEG signals recorded during smelling of the present odors. It has been also established that gamma-band activity of the brain is highly associated with olfaction.

Identification of Anisomerous Motor Imagery EEG Signals Based on Complex Algorithms.

Motor imagery (MI) electroencephalograph (EEG) signals are widely applied in brain-computer interface (BCI). However, classified MI states are limited, and their classification accuracy rates are low because of the characteristics of nonlinearity and nonstationarity. This study proposes a novel MI pattern recognition system that is based on complex algorithms for classifying MI EEG signals. In electrooculogram (EOG) artifact preprocessing, band-pass filtering is performed to obtain the frequency band of MI-related signals, and then, canonical correlation analysis (CCA) combined with wavelet threshold denoising (WTD) is used for EOG artifact preprocessing. We propose a regularized common spatial pattern (R-CSP) algorithm for EEG feature extraction by incorporating the principle of generic learning. A new classifier combining the K-nearest neighbor (KNN) and support vector machine (SVM) approaches is used to classify four anisomerous states, namely, imaginary movements with the left hand, right foot, and right shoulder and the resting state. The highest classification accuracy rate is 92.5%, and the average classification accuracy rate is 87%. The proposed complex algorithm identification method can significantly improve the identification rate of the minority samples and the overall classification performance.

A Parallel Adaboost-Backpropagation Neural Network for Massive Image Dataset Classification.

Image classification uses computers to simulate human understanding and cognition of images by automatically categorizing images. This study proposes a faster image classification approach that parallelizes the traditional Adaboost-Backpropagation (BP) neural network using the MapReduce parallel programming model. First, we construct a strong classifier by assembling the outputs of 15 BP neural networks (which are individually regarded as weak classifiers) based on the Adaboost algorithm. Second, we design Map and Reduce tasks for both the parallel Adaboost-BP neural network and the feature extraction algorithm. Finally, we establish an automated classification model by building a Hadoop cluster. We use the Pascal VOC2007 and Caltech256 datasets to train and test the classification model. The results are superior to those obtained using traditional Adaboost-BP neural network or parallel BP neural network approaches. Our approach increased the average classification accuracy rate by approximately 14.5% and 26.0% compared to the traditional Adaboost-BP neural network and parallel BP neural network, respectively. Furthermore, the proposed approach requires less computation time and scales very well as evaluated by speedup, sizeup and scaleup. The proposed approach may provide a foundation for automated large-scale image classification and demonstrates practical value.

Classification Recognition Algorithm Based on Strong Association Rule Optimization of Neural Network

Feature selection of text is one of the basic matters for intelligent classification of text. Textual feature generating algorithm adopts weighted textual vector space model generally at present. This model uses BP network evaluation function to calculate weight value of single feature and textual feature redundancy generated in this algorithm is high generally. For this problem, a textual feature generating algorithm based on clustering weighting is adopted. This new method conducts initial weighted treatment for feature candidate set first of all and then conducts further weighted treatment of features through semantic and information entropy and it removes redundancy features with features clustering at last. Experiment shows that the average classification accuracy rate of this algorithm is about 5% higher than that of traditional BP network algorithm.

SEMG-Based Identification of Hand Motion Commands Using Wavelet Neural Network Combined With Discrete Wavelet Transform

Surface electromyogram (sEMG) signals can be applied in medical, rehabilitation, robotic, and industrial fields. As a typical application, a myoelectric prosthetic hand is controlled by the sEMG signals of the amputee’s residual muscles. To improve the dexterity of the myoelectric prosthetic hand, additional hand motion commands need to be classified. The more sEMG sensors are used, the more hand motion commands can be classified. However, the amputee’s residual muscles are limited. In order to improve the practicability of the myoelectric prosthetic hand, it is critical to investigate the effective pattern recognition algorithms to deal with the sEMG signals detected by fewer sensors, while identifying as many hand motion commands as possible. Current pattern recognition algorithms for sEMG signals are challenged by limited recognition patterns and unsteady classification accuracy rates. To solve these dilemmas, we employed discrete wavelet transform (DWT) and wavelet neural network (WNN) algorithms to improve the pattern recognition effects of sEMG signals. In addition, the back propagation and gradient descent algorithms were utilized to train WNN. In this work, we only used three sEMG sensors to classify and recognize six kinds of hand motion commands. The maximum identification accuracy rate is 100%, and an average classification accuracy rate of the proposed WNN is 94.67%, which is substantially better than the artificial neural network (ANN) algorithm.

A classifier fusion approach to osteoporosis prediction for women in Taiwan

Osteoporosis is a progressive bone disease which will increase the risk of fracture. Related studies show that women are much more likely to get osteoporosis than men. In Taiwan, bone mineral density examination (BMD) is not included in basic health examination and needs additional payments. To reduce the healthcare burden on the government and the public, this study proposed a classifier fusion approach to develop an osteoporosis prediction model based on the past health examination data (including the basic health examination and paid BMD examination) for women in Taiwan. The proposed classifier fusion approach integrates genetic algorithm (GA), classifier and voting scheme into a GA-based ensemble classifier (EC) with the consideration of feature selection, parameter optimization and multidimensional forecasting. The results show that GA-based EC with an average classification accuracy rate of 70.43% outperforms individual classifier. Based on prediction results, the high-risk patients will be recommended to subsequently BMD examinations.

Brain Computer Interface system based on indoor semi-autonomous navigation and motor imagery for Unmanned Aerial Vehicle control

This paper proposes a non-invasive Electroencephalogram (EEG)-based Brain Computer Interface (BCI) system to achieve the easy-to-use and stable control of a low speed Unmanned Aerial Vehicle (UAV) for indoor target searching. The BCI system for UAV control consists of two main subsystems responsible for decision and semi-autonomous navigation. The decision subsystem is established based on the analysis of motor imagery (MI) EEG. The improved cross-correlation method (CC) is used to accomplish the MI feature extraction and the logistic regression method (LR) is employed to complete the MI feature classification and decision. The average classification accuracy rate of the BCI system reaches to 94.36%. The semi-autonomous navigation subsystem is utilized to avoid obstacles automatically for UAV and provide feasible directions for decision subsystem. The actual indoor target searching experiment is carried out to verify the performance of this BCI system. The experiment validates the feasibility and effectiveness of this BCI system for low speed UAV control by using MI and semi-autonomous navigation.

Classification of Evoked Emotions Using an Artificial Neural Network Based on Single, Short-Term Physiological Signals

Most automated analysis methods related to biosignalbased human Emotions collect their data using multiple physiological signals, long-term physiological signals, or both. However, this restricts their ability to identify Emotions in an efficient manner. This study classifies evoked Emotions based on two types of single, short-term physiological signals: electrocardiograms (ECGs) and galvanic skin responses (GSRs) respectively. Estimated recognition times are also recorded and analyzed. First, we perform experiments using film excerpts selected to elicit target Emotions that include anger, grief, fear, happiness, and calmness; ECG and GSR signals are collected during these experiments. Next, a wavelet transform is applied to process the truncated ECG data, and a Butterworth filter is applied to process the truncated GSR signals, in order to extract the required features. Finally, the five different Emotion types are classified by employing an artificial neural network (ANN) based on the two signals. Average classification accuracy rates of 89.14% and 82.29% were achieved in the experiments using ECG data and GSR data, respectively. In addition, the total time required for feature extraction and emotional classification did not exceed 0.15 s for either ECG or GSR signals.

Karhunen-Loeve Transform and Sparse Representation Based Plant Leaf Disease Recognition

To improve the classification accuracy rate of apple leaf disease images and solve the problem of dimension redundancy in feature extraction, Karhunen-Loeve (K-L) transform and sparse representation are applied to apple leaf disease recognition. Firstly 9 color features and 8 texture features of disease leaf images are extracted and taken as feature vectors after dimensionality reduction by the K-L transform. Then, for each of apple mosaic virus, apple rust and apple alternaria leaf spot, 40 apple leaf images are selected as the training samples, whose feature vectors are made up of the dictionary of the sparse representation, respectively. Each testing sample is classified into the class with the minimal residual. The identifying results using the proposed method are analyzed and compared with those of the Support Vector Machine (SVM) and original sparse representation method. The average classification accuracy rate of the proposed method is 94.18 %, which confirms its good robustness. In addition, the proposed method not only improves the plant leaf disease classification accuracy but also solves the redundancy problem of the extracted features. DOI : http://dx.doi.org/10.11591/telkomnika.v12i2.3632

Detection of Financial Statement Fraud Using Evolutionary Algorithms

ABSTRACTIn this paper, we use a Genetic Algorithm (GA) and MARLEDA—a modern Estimation of Distribution Algorithm (EDA)—to evolve and train several fuzzy rule-based classifiers (FRBCs) to detect patterns of financial statement fraud. We find that both GA and MARLEDA demonstrate a better ability to classify unseen corporate data observations than those of a traditional logistic regression model, and provide validity for detecting financial statement fraud with Evolutionary Algorithms (EAs) and FRBCs. Using ten-fold cross-validation, the GA and MARLEDA yield average training classification accuracy rates of 75.47 percent and 74.26 percent, respectively, and average validation accuracy rates of 63.75 percent and 64.46 percent, respectively.

Generating Interpretable Fuzzy Systems for Classification Problems

En este artículo se presenta un nuevo método para generar sistemas difusos interpretables, a partir de datos experimentalesde entrada y salida, para resolver problemas de clasificación. En la partición antecedente se emplean conjuntos triangulares con interpolación de 0.5 lo cual evita la presencia de solapamientos complejos que suceden en otros métodos. Los consecuentes, tipo Singleton, son generados por la proyección de los valores modales de cada función de membresía triangular en el espacio de salida y se emplea el método de mínimos cuadrados para el ajuste de los consecuentes. El método propuesto consigue una mayor precisión que la alcanzada con los métodos actuales existentes, empleando un número reducido de reglas y parámetros y sin sacrificar la interpretabilidad del modelo difuso. El enfoque propuesto es aplicado a dos problemas clásicos de clasificación: el Wisconsin Breast Cancer (WBC) y el Iris Data Classification Problem, para mostrar las ventajas del método y comparar los resultados con los alcanzados por otros investigadores.

Fuzzy classification systems based on fuzzy information gain measures

In this paper, we present a new method for handling classification problems using a new fuzzy information gain measure. Based on the proposed fuzzy information gain measure, we propose an algorithm for constructing membership functions, calculating the class degree of each subset of training instances with respect to each class and calculating the fuzzy entropy of each subset of training instances. Based on the constructed membership function of each fuzzy set of each feature, the obtained class degree of each subset of training instances with respect to each class and the obtained fuzzy entropy of each subset of training instances, we propose an evaluating function for classifying testing instances. The proposed method gets higher average classification accuracy rates than the methods presented in [John, G. H., & Langley, P. (1995). Estimating continuous distributions in Bayesian classifiers. In Proceedings of the 11th conference on uncertainty in artificial intelligence, Montreal, Canada (pp. 338–345); Platt, J. C. (1999). Using analytic QP and sparseness to speed training of support vector machines. In Proceedings of the 13th annual conference on neural information processing systems, Denver, Colorado (pp. 557–563); Quinlan, J. R. (1993). C4.5: Programs for machine learning. San Francisco: Morgan Kaufmann].

Frog classification using machine learning techniques

An automatic frog sound identification system is developed in this work to provide the public to easily consult online. The sound samples are first properly segmented into syllables. Then three features, spectral centroid, signal bandwidth and threshold-crossing rate, are extracted to serve as the parameters for the frog sound classification. Two well-known classifiers, kNN and SVM, are adopted to recognize the frog species based on the three extracted features. The experimental results show that the average classification accuracy rate can be up to 89.05% and 90.30% for kNN and SVM classifiers, respectively. The effectiveness of the proposed on-line recognition system is thus verified.

Generating fuzzy rules from training instances for fuzzy classification systems

In recent years, many methods have been proposed to generate fuzzy rules from training instances for handling the Iris data classification problem. In this paper, we present a new method to generate fuzzy rules from training instances for dealing with the Iris data classification problem based on the attribute threshold value @a, the classification threshold value @b and the level threshold value @c, where @a@?[0,1], @b@?[0,1] and @c@?[0,1]. The proposed method gets a higher average classification accuracy rate than the existing methods.

Feature subset selection based on fuzzy entropy measures for handling classification problems

In this paper, we present a new method for dealing with feature subset selection based on fuzzy entropy measures for handling classification problems. First, we discretize numeric features to construct the membership function of each fuzzy set of a feature. Then, we select the feature subset based on the proposed fuzzy entropy measure focusing on boundary samples. The proposed method can select relevant features to get higher average classification accuracy rates than the ones selected by the MIFS method (Battiti, R. in IEEE Trans. Neural Netw. 5(4):537---550, 1994), the FQI method (De, R.K., et al. in Neural Netw. 12(10):1429---1455, 1999), the OFEI method, Dong-and-Kothari's method (Dong, M., Kothari, R. in Pattern Recognit. Lett. 24(9):1215---1225, 2003) and the OFFSS method (Tsang, E.C.C., et al. in IEEE Trans. Fuzzy Syst. 11(2):202---213, 2003).

A NEW METHOD TO CONSTRUCT MEMBERSHIP FUNCTIONS AND GENERATE WEIGHTED FUZZY RULES FROM TRAINING INSTANCES

Fuzzy classification systems are important applications of the fuzzy set theory. In order to design a fuzzy classification system, it is an important task to construct the membership function of each attribute and generate fuzzy rules that are suitable for handling a specific classification problem. In this paper, we propose a new method to construct the membership function of each attribute and generate weighted fuzzy rules from training instances for handling fuzzy classification problems. The proposed method can construct membership functions and generate weighted fuzzy rules without any human experts' intervention. It can get a higher average classification accuracy rate and generate fewer fuzzy rules than the existing methods.