Classification and detection using hidden Markov model‐support vector machine algorithm based on optimal colour space selection for blood images

In this paper a novel approach based on multiple color channels is proposed for multi-class fruit detection. Multi-feature fusion of global color histogram, Local Binary Pattern (LBP), Histogram of Oriented Gradient (HOG), and LBP based on Gabor wavelets (GaborLBP) is utilized to improve fruit recognition. These four features are extracted on multiple color channels. To select the optimal features of HOG, LBP and GaborLBP in each color channel, the optimal block is selected according to the cross validation accuracy. The optimal color channel fusion is optimized by a simple optimal color channel selection framework proposed in this paper. A multi-class fruit classifier is trained using the optimal color channel fusion and LibSVM. Besides, the proposed method is evaluated on a fruit dataset including 5 classes of fruit and 1778 test images. The experiments show that the proposed multi-class fruit detection method is effective. This proposed method has produced 0.156 miss rate at 0.5427 false positives per image (FPPI) and high average precision (0.8135), and can detect multi-class and multiple fruits in a variety of backgrounds, locations, angles, and sizes.

Bond rating is regarded as an important event for measuring financial risk of companies and for determining the investment returns of investors. As a result, it has been a popular research topic for researchers to predict companies' credit ratings by applying statistical and machine learning techniques. The statistical techniques, including multiple regression, multiple discriminant analysis (MDA), logistic models (LOGIT), and probit analysis, have been traditionally used in bond rating. However, one major drawback is that it should be based on strict assumptions. Such strict assumptions include linearity, normality, independence among predictor variables and pre-existing functional forms relating the criterion variablesand the predictor variables. Those strict assumptions of traditional statistics have limited their application to the real world. Machine learning techniques also used in bond rating prediction models include decision trees (DT), neural networks (NN), and Support Vector Machine (SVM). Especially, SVM is recognized as a new and promising classification and regression analysis method. SVM learns a separating hyperplane that can maximize the margin between two categories. SVM is simple enough to be analyzed mathematical, and leads to high performance in practical applications. SVM implements the structuralrisk minimization principle and searches to minimize an upper bound of the generalization error. In addition, the solution of SVM may be a global optimum and thus, overfitting is unlikely to occur with SVM. In addition, SVM does not require too many data sample for training since it builds prediction models by only using some representative sample near the boundaries called support vectors. A number of experimental researches have indicated that SVM has been successfully applied in a variety of pattern recognition fields. However, there are three major drawbacks that can be potential causes for degrading SVM's performance. First, SVM is originally proposed for solving binary-class classification problems. Methods for combining SVMs for multi-class classification such as One-Against-One, One-Against-All have been proposed, but they do not improve the performance in multi-class classification problem as much as SVM for binary-class classification. Second, approximation algorithms (e.g. decomposition methods, sequential minimal optimization algorithm) could be used for effective multi-class computation to reduce computation time, but it could deteriorate classification performance. Third, the difficulty in multi-class prediction problems is in data imbalance problem that can occur when the number of instances in one class greatly outnumbers the number of instances in the other class. Such data sets often cause a default classifier to be built due to skewed boundary and thus the reduction in the classification accuracy of such a classifier. SVM ensemble learning is one of machine learning methods to cope with the above drawbacks. Ensemble learning is a method for improving the performance of classification and prediction algorithms. AdaBoost is one of the widely used ensemble learning techniques. It constructs a composite classifier by sequentially training classifiers while increasing weight on the misclassified observations through iterations. The observations that are incorrectly predicted by previous classifiers are chosen more often than examples that are correctly predicted. Thus Boosting attempts to produce new classifiers that are better able to predict examples for which the current ensemble's performance is poor. In this way, it can reinforce the training of the misclassified observations of the minority class. This paper proposes a multiclass Geometric Mean-based Boosting (MGM-Boost) to resolve multiclass prediction problem. Since MGM-Boost introduces the notion of geometric mean into AdaBoost, it can perform learning process considering the geometric mean-based accuracy and errors of multiclass. This study applies MGM-Boost to the real-world bond rating case for Korean companies to examine the feasibility of MGM-Boost. 10-fold cross validations for threetimes with different random seeds are performed in order to ensure that the comparison among three different classifiers does not happen by chance. For each of 10-fold cross validation, the entire data set is first partitioned into tenequal-sized sets, and then each set is in turn used as the test set while the classifier trains on the other nine sets. That is, cross-validated folds have been tested independently of each algorithm. Through these steps, we have obtained the results for classifiers on each of the 30 experiments. In the comparison of arithmetic mean-based prediction accuracy between individual classifiers, MGM-Boost (52.95%) shows higher prediction accuracy than both AdaBoost (51.69%) and SVM (49.47%). MGM-Boost (28.12%) also shows the higher prediction accuracy than AdaBoost (24.65%) and SVM (15.42%)in terms of geometric mean-based prediction accuracy. T-test is used to examine whether the performance of each classifiers for 30 folds is significantly different. The results indicate that performance of MGM-Boost is significantly different from AdaBoost and SVM classifiers at 1% level. These results mean that MGM-Boost can provide robust and stable solutions to multi-classproblems such as bond rating.

BackgroundA nanopore detector has a nanometer-scale trans-membrane channel across which a potential difference is established, resulting in an ionic current through the channel in the pA-nA range. A distinctive channel current blockade signal is created as individually "captured" DNA molecules interact with the channel and modulate the channel's ionic current. The nanopore detector is sensitive enough that nearly identical DNA molecules can be classified with very high accuracy using machine learning techniques such as Hidden Markov Models (HMMs) and Support Vector Machines (SVMs).ResultsA non-standard implementation of an HMM, emission inversion, is used for improved classification. Additional features are considered for the feature vector employed by the SVM for classification as well: The addition of a single feature representing spike density is shown to notably improve classification results. Another, much larger, feature set expansion was studied (2500 additional features instead of 1), deriving from including all the HMM's transition probabilities. The expanded features can introduce redundant, noisy information (as well as diagnostic information) into the current feature set, and thus degrade classification performance. A hybrid Adaptive Boosting approach was used for feature selection to alleviate this problem.ConclusionThe methods shown here, for more informed feature extraction, improve both classification and provide biologists and chemists with tools for obtaining a better understanding of the kinetic properties of molecules of interest.

The classification of histopathological imagery has garnered significant interest among researchers in the last decade due to the valuable outcome that could be obtained from classifying such microscopic fractions. This would significantly contribute to examining biological interactions. To do so, researchers in the literature have employed various machine-learning classification algorithms. However, the key to success for a precise classification task lies in utilizing an appropriate set of features with proper color space channels that can extract important characteristics from the histopathological images. However, the literature shows a limited feature extraction method with limited color space channel utilization. The accuracy of classification is significantly influenced by the color channels. This study aims to extend feature learning by using a wide range of feature extraction methods and different employs distinct color channels to categorize histopathological imagery. It utilizes two benchmark datasets pertinent to the imagery of breast and prostate cancer for the study. Additionally, the study incorporated a series of pre-processing procedures, such as segmenting the images and extracting salient features. Image segmentation in this research was conducted using four distinct methodologies, encompassing Lumen, Nuclei, Cytoplasm, and Stroma. The reason behind selecting such feature extraction methods and color channels is their popularity and differences, which ensure diversity. Finally, the study utilized SVM-RFE to select features and classify images. The assessment employed metrics such as sensitivity, f1-score, and accuracy. The empirical findings demonstrate that the RGB color space yielded superior performance particularly on the sensitivity evaluation metrics across both datasets, underscoring RGB’s efficacy in classifying histopathological images.

This paper describes a color space optimized for processing images of uniformly colored objects. Standard RGB color space is an ineffective space for this type of processing due to the high cross-correlation between color channels. The proposed /spl beta//spl delta//spl epsiv/ color space establishes a primary axis containing the expected color range of the object. The second and third coordinates contain magnitude and chromaticity of any deviation from the expected color. The separation into application specific coordinates is the main advantage of the new color space. The new color space has been used in two industrial inspection systems and has been implemented in hardware.

The current classification algorithms have weak fault-tolerance. In order to solve problem, a multiple support vector machines method, called Upper preferred Multiple Directed Acyclic Graph Support Vector Machines (UMDAG-SVMs), is proposed. Firstly, we present least squares projection twin support vector machine (LSPTSVM) with confidence-degree for generating binary classifiers. It uses idea that when confidence-degree outputted from node in directed graph, is below threshold, decision-making process will go on along with two branches of node at same time., which strengthens algorithm's fault-tolerance. In order to select parameters of algorithm, we use genetic algorithm to select these parameters. Secondly, according to minimal hypersphere distance, and known principle the upper-level classifiers bring up better performance of classification in DAG-SVMs , we present a new classification algorithm, called UMDAG-SVMs. This algorithm has two advantages of strong fault-tolerance and high classification accuracy. Finally, we make experiments to test performance of algorithm. Experimental results in public datasets show that our UMDAG-SVMs has comparable classification accuracy to that other algorithms but with remarkable less computation.

In this paper, we study to name a color of an object in the given image, where the number of color names is 11 English basic color names, including pink, purple, red, orange, yellow, green, blue, gray, brown, black, and white. Our goal is to apply machine learning algorithms for assigning a color name to a principal object in images. Here, we evaluate three basic classification algorithms, namely Support Vector Machine (SVM), naive Bayes, and k-Nearest Neighbor (k-NN). The feature vector of the object in an image is the pixel intensities represented in different color spaces including RGB, HSV, Lab and CYMK. We randomly collect the training pixel intensities with their corresponding color names from the images obtained from the Internet. We then evaluate three classification algorithms using the e-Bay dataset. The experimental results show that the linear SVM with Lab color space is the best choice for solving this task.

Smoothing methods, extensively used for solving important mathematical programming problems and applications, are applied here to generate and solve an unconstrained smooth reformulation of the support vector machine for pattern classification using a completely arbitrary kernel. We term such reformulation a smooth support vector machine (SSVM). A fast Newton–Armijo algorithm for solving the SSVM converges globally and quadratically. Numerical results and comparisons are given to demonstrate the effectiveness and speed of the algorithm. On six publicly available datasets, tenfold cross validation correctness of SSVM was the highest compared with four other methods as well as the fastest. On larger problems, SSVM was comparable or faster than SVMlight (T. Joachims, in Advances in Kernel Methods—Support Vector Learning, MIT Press: Cambridge, MA, 1999), SOR (O.L. Mangasarian and David R. Musicant, IEEE Transactions on Neural Networks, vol. 10, pp. 1032–1037, 1999) and SMO (J. Platt, in Advances in Kernel Methods—Support Vector Learning, MIT Press: Cambridge, MA, 1999). SSVM can also generate a highly nonlinear separating surface such as a checkerboard.

The inclusion in workplaces of video display terminals has brought multiple benefits for the organization of work. Nevertheless, it also implies a series of risks for the health of the workers, since it can cause ocular and visual disorders, among others. In this research, a group of eye and vision-related problems associated with prolonged computer use (known as computer vision syndrome) are studied. The aim is to select the characteristics of the subject that are most relevant for the occurrence of this syndrome, and then, to develop a classification model for its prediction. The estimate of this problem is made by means of support vector machines for classification. This machine learning technique will be trained with the support of a genetic algorithm. This provides the training of the support vector machine with different patterns of parameters, improving its performance. The model performance is verified in terms of the area under the ROC curve, which leads to a model with high accuracy in the classification of the syndrome.

To acquire the fruit region of green apple image in close color more completely, this work designed a method of region extraction before fruit region combination. Firstly, channel images in every color space were analyzed to determine appropriate color channel images: images of G and b color channel in RGB and Lab color spaces, respectively. Then, these images were segmented by K-means clustering to obtain segmented images of G and b color channel. Post-processing was conducted to different segmentation results. For example, open operation and micro-region elimination were conducted to segmented image of G color channel. The same processing was conducted to b color channel image after combining with above processed G color channel image and filling holes. Therefore, the target fruit region can be acquired completely. The validity of this method was verified by experiment.

Lesion localization is the first and most important step in the development of a melanoma detection system. Skin lesions show distinct color variations as they evolve from benign to malignant one. The use of different color spaces can provide the best discriminative information of lesions embedded in a particular color channel and used in efficient segmentation of a digital image. A simple and efficient method for the automatic segmentation of clinical images using colorspace analysis and improved binary thresholding algorithm is proposed. Clinical images taken from normal mobile cameras have the inherent problem of improperly illumined background compared to dermoscopic images. The proposed algorithm corrects these constraints with initial pre-processing steps of adjustments of clinical image. The algorithm has been evaluated using 175 images from different online public databases. The efficiency of the overall segmentation process is carried out by calculation of similarity matrices for segmented lesions from each color channel used. The framework is able to differentiate and extract the cancerous lesion from the background skin with around 94% accuracy with the preferred choice of color channel.

The diagnosis of cardiac dysfunctions requires the analysis of long-term ECG signal recordings, often containing hundreds to thousands of heartbeats. The purpose of this work is to propose a diagnostic system for modelling and classification of heartbeat, by use of time features and Support vector machines (SVM) classification algorithm. Neural Networks learning allow us to select a features of each heart beat on the basis of Generalized Orthogonal Forward Regression (GOFR) algorithm and a library of 132 Gaussians with different standard deviations and different means, each beat is represented by five Gaussians with different amplitudes. The parameters of this system are determined and its performance is evaluated for the MIT-BIH arrhythmia database. For a database of 364 normal heartbeats and 1148 abnormal heartbeats, we apply the SVM algorithm with Radial Basis Function kernel. Our results demonstrate that the testing performance of the neural network and SVM diagnostic system is found to be very satisfactory with a recognition rate of 99.67%.

Speech emotion recognition as a significant part has become a challenge to artificial emotion. It is particularly difficult to recognize emotion independent of the person concentrating on the speech channel. In the paper, an integrated system of hidden Markov model (HMM) and support vector machine (SVM), which combining advantages on capability to dynamic time warping of HMM and pattern recognition of SVM, has been proposed to implement speaker independent emotion classification. Firstly, all emotions are divided into two groups by SVM. Then, HMMs are used to discriminate emotions from each group. For a more robust estimation, we also combine four HMMs classifiers into a system. The recognition result of the fusion system has been compared with the isolated HMMs using Mandarin database. Experimental results demonstrate that comparing with the method based on only HMMs, the proposed system is more effective and the average recognition rate reaches 76.1% when speaker is independent.

Support vector machine for classification based on fuzzy training data

Reconstruction of the nipple-areola complex is the final process in breast reconstruction. Local flaps and paramedical pigmentation is one of the major procedures for this. However, fading after paramedical pigmentation leads to a color difference between the selected pigment and its color in the skin. The aim of this study is to make a proposition in color choice of paramedical pigmentation for nipple-areola complex. Our research focused on investigating the color changes over time after unilateral nipple-areola complex reconstruction using paramedical pigmentation in 25 patients to propose suitable color selections. We measured the color by spectrometer and conducted comparisons using the hue, saturation, and value (HSV) color space and the color space defined by the Commission International de L'eclairage based on one channel for luminance (lightness) (L) and two color channels (a and b) (L*a*b*). A comparison of the hue, value, and saturation of the reconstructed areola compared to the normal areolae was conducted using HSV color space; the value and saturation were satisfactory after 3months and beyond, but the reconstructed areola tended to have stronger red hues. The color difference (ΔE00) calculated in L*a*b* color space showed slow fading after the scab was peeled off. This result indicates that a color with less redness and more yellowness, particularly 4-5 degrees of yellowness on the color wheel, than the normal side is the most appropriate color selection for this technique. This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .