Multi-class Pattern Recognition Research Articles

TCN: Pioneering Topological-Based Convolutional Networks for Planetary Terrain Learning

Implementations of artificial intelligence (AI) based on deep learning (DL) have proven to be highly successful in many domains, from biomedical imaging to natural language processing, but are still rarely applied in the space industry, particularly for onboard learning of planetary surfaces. In this project, we discuss the utility and limitations of DL, enhanced with topological footprints of the sensed objects, for multi-class classification of planetary surface patterns, in conjunction with tactile and embedded sensing in rover exploratory missions. We consider a Topological Convolutional Network (TCN) model with a persistence-based attention mechanism for supervised classification of various landforms. We study TCN's performance on the Barefoot surface pattern dataset, a novel surface pressure dataset from a prototype tactile rover wheel, known as the Barefoot Rover tactile wheel. Multi-class pattern recognition in the Barefoot data has neither been ever tackled before with DL nor assessed with topological methods. We provide insights into advantages and restrictions of topological DL as the early-stage concept for onboard learning and planetary exploration.

A Fast-Efficient Multi Class Pattern Recognition Method

This work presents a novel method for multi class pattern recognition. The feature space is classified with minimum hardware complexity and maximum speed using straight lines, circles, parabolas etc. RK algorithm-based devices (RKD) and mathematical functional blocks classify the feature space very rapidly after learning pattern classification with a fewer numbers of training sets compared to other statistical and artificial neural network (ANN) methods. RKDs are self-learning and fast responding devices and which manipulate a single variable at a time. The RK algorithm is used for learning the range of a variable. A set of sample variable and their corresponding responds are given for learning. The mathematical functional blocks manipulate one or more variables or attributes to perform mathematical functions and the outputs of these blocks are fed to RKDs. Finally, the RKDs perform the classification functions. The classification using straight lines or curves depends upon the mathematical functional block.

Convex hulls and convex separable sets in the multiclass pattern recognition

Convex hulls and convex separable sets in the multiclass pattern recognition

Machine Learning for Wind Turbine Blades Maintenance Management

Delamination in Wind Turbine Blades (WTB) is a common structural problem that can generate large costs. Delamination is the separation of layers of a composite material, which produces points of stress concentration. These points suffer greater traction and compression forces in working conditions, and they can trigger cracks, and partial or total breakage of the blade. Early detection of delamination is crucial for the prevention of breakages and downtime. The main novelty presented in this paper has been to apply an approach for detecting and diagnosing the delamination WTB. The approach is based on signal processing of guided waves, and multiclass pattern recognition using machine learning. Delamination was induced in the WTB to check the accuracy of the approach. The signal is denoised by wavelet transform. The autoregressive Yule–Walker model is employed for feature extraction, and Akaike’s information criterion method for feature selection. The classifiers are quadratic discriminant analysis, k-nearest neighbors, decision trees, and neural network multilayer perceptron. The confusion matrix is employed to evaluate the classification, especially the receiver operating characteristic analysis by: recall, specificity, precision, and F-score.

Time–frequency envelope analysis-based sub-band selection and probabilistic support vector machines for multi-fault diagnosis of low-speed bearings

This paper proposes a highly reliable multi-fault diagnosis scheme for low-speed rolling element bearings using an effective time–frequency envelope analysis and a Bayesian inference based one-against-all support vector machines (probabilistic-OAASVM) classifier. The proposed method first performs a wavelet packet transform based envelope analysis on an acoustic emission signal to select sub-bands of the signal that contain the most intrinsic and pertinent information about the defects. Frequency- and time-domain fault features are extracted only from selected sub-bands for fault classification. Traditional one-against-all SVMs (OAASVM), a widely used multi-class pattern recognition technique, employ an arbitrary combination of a series of binary classifiers yielding overlapped feature spaces, where a data sample can be unclassifiable. To address this limitation, we formulate the feature space of OAASVM as an appropriate Gaussian process prior (GPP) and interpret OAASVM results as a posterior probability estimation procedure using Bayesian inference under this GPP. The efficacy of the proposed probabilistic-OAASVM classifier is verified for low-speed rolling element bearings under various conditions. Experimental results indicate that the proposed method outperforms the state-of-the-art algorithms for multi-fault classification of low-speed bearings, yielding a 4.95–20.67% improvement in the average classification accuracy.

AN ARTIFICIAL NEURAL NETWORK MODEL FOR OUTCOME PREDICTION IN GASTRIC CANCER PATIENTS

Multiclass pattern recognition is a problem of building a system that accurately maps an input feature space to an output space of more than two pattern classes. K-class pattern classification can be implemented in a single neural network with K output nodes. Such a model can be extended to make predictions about patients’ probability of survival over time. This paper proposes a multiple time-point ANN model for predicting the probability of survival at different time intervals for patients with gastric cancer. More specifically, survival is modeled using a multiple-output ANN, with a structure modulated to produce different values as the probability of survival for each time interval. The model’s performance in outcome prediction is investigated with a real gastric cancer data set.

Многоклассовое распознавание образов в пространстве представлений с многоуровневым разрешением

Многоклассовое распознавание образов в пространстве представлений с многоуровневым разрешением

A probabilistic multi-class classifier for structural health monitoring

In this paper, a probabilistic multi-class pattern recognition algorithm is developed for damage monitoring of smart structures. As these structures can face damages of different severities located in various positions, multi-class classifiers are needed. We propose an original support vector machine (SVM) multi-class clustering algorithm that is based on a probabilistic decision tree (PDT) that produces a posteriori probabilities associated with damage existence, location and severity. The PDT is built by iteratively subdividing the surface and thus takes into account the structure geometry. The proposed algorithm is very appealing as it combines both the computational efficiency of tree architectures and the SVMs classification accuracy. Damage sensitive features are computed using an active approach based on the permanent emission of non-resonant Lamb waves into the structure and on the recognition of amplitude disturbed diffraction patterns. The effectiveness of this algorithm is illustrated experimentally on a composite plate instrumented with piezoelectric elements.

MULTICLASS PATTERN RECOGNITION OF THE GLEASON SCORE OF PROSTATIC CARCINOMAS USING METHODS OF SPATIAL STATISTICS

The Gleason score of a prostatic carcinoma is generally considered as one of the most important prognostic parameters of this tumour type. In the present study, it was attempted to study the relation between the Gleason score and objective data of spatial statistics, and to predict this score from such data. For this purpose, 25 T1 incidental prostatic carcinomas, 50 pT2N0, and 28 pT3N0 prostatic adenocarcinomas were characterized by a histological texture analysis based on principles of spatial statistics. On sectional images, progression from low grade to high grade prostatic cancer in terms of the Gleason score is correlated with complex changes of the epithelial cells and their lumina with respect to their area, boundary length and Euler number per unit area. The central finding was a highly significant negative correlation between the Gleason score and the Euler number of the epithelial cell phase per unit area. The Gleason score of all individual cases was predicted from the spatial statistical variables by multivariate linear regression. This approach means to perform a multiclass pattern recognition, as opposed to the usual problem of binary pattern recognition. A prediction was considered as acceptable when its deviation from the human classification was no more than 1 point. This was achieved in 79 of these 103 cases when only the Euler number density was used as predictor variable. The accuracy could be risen slightly to 84 of the 103 cases, when 7 input variables were used for prediction of the Gleason score, which means an accuracy of 81.5%.

Multi-class pattern recognition with the group match competition method

High recognition rate of multi-class pattern recognition is difficult to obtain, especially when the feature extraction is not ideal. So achieving a high recognition rate with nonideal feature extraction makes great sense. Based on the early study of multiple-set-compete method (MSCM) and the algorithm of result reliability, this article proposes a new recognition method named group match competition method (GMCM). The GMCM precedes the early work with a larger scope of applications. Early work which the authors did can only deal with the recognition work with the number of powers of 2 classes, while GMCM can cope with the classes of any number. This article further illustrates the MSCM, the algorithm of result reliability and their functions in the GMCM. Three sets of cases are demonstrated and formats of grouping are discussed. The recognition results show that the GMCM is a robust method and it is capable of achieving the high recognition rate of the multi-class pattern recognition.

Automated, high accuracy classification of Parkinsonian disorders: a pattern recognition approach.

Progressive supranuclear palsy (PSP), multiple system atrophy (MSA) and idiopathic Parkinson’s disease (IPD) can be clinically indistinguishable, especially in the early stages, despite distinct patterns of molecular pathology. Structural neuroimaging holds promise for providing objective biomarkers for discriminating these diseases at the single subject level but all studies to date have reported incomplete separation of disease groups. In this study, we employed multi-class pattern recognition to assess the value of anatomical patterns derived from a widely available structural neuroimaging sequence for automated classification of these disorders. To achieve this, 17 patients with PSP, 14 with IPD and 19 with MSA were scanned using structural MRI along with 19 healthy controls (HCs). An advanced probabilistic pattern recognition approach was employed to evaluate the diagnostic value of several pre-defined anatomical patterns for discriminating the disorders, including: (i) a subcortical motor network; (ii) each of its component regions and (iii) the whole brain. All disease groups could be discriminated simultaneously with high accuracy using the subcortical motor network. The region providing the most accurate predictions overall was the midbrain/brainstem, which discriminated all disease groups from one another and from HCs. The subcortical network also produced more accurate predictions than the whole brain and all of its constituent regions. PSP was accurately predicted from the midbrain/brainstem, cerebellum and all basal ganglia compartments; MSA from the midbrain/brainstem and cerebellum and IPD from the midbrain/brainstem only. This study demonstrates that automated analysis of structural MRI can accurately predict diagnosis in individual patients with Parkinsonian disorders, and identifies distinct patterns of regional atrophy particularly useful for this process.

The application of improved NeuroEvolution of Augmenting Topologies neural network in Marcellus Shale lithofacies prediction

The organic-rich Marcellus Shale was deposited in a foreland basin during Middle Devonian. In terms of mineral composition and organic matter richness, we define seven mudrock lithofacies: three organic-rich lithofacies and four organic-poor lithofacies. The 3D lithofacies model is very helpful to determine geologic and engineering sweet spots, and consequently useful for designing horizontal well trajectories and stimulation strategies. The NeuroEvolution of Augmenting Topologies (NEAT) is relatively new idea in the design of neural networks, and shed light on classification (i.e., Marcellus Shale lithofacies prediction). We have successfully enhanced the capability and efficiency of NEAT in three aspects. First, we introduced two new attributes of node gene, the node location and recurrent connection (RCC), to increase the calculation efficiency. Second, we evolved the population size from an initial small value to big, instead of using the constant value, which saves time and computer memory, especially for complex learning tasks. Third, in multiclass pattern recognition problems, we combined feature selection of input variables and modular neural network to automatically select input variables and optimize network topology for each binary classifier. These improvements were tested and verified by true if an odd number of its arguments are true and false otherwise (XOR) experiments, and were powerful for classification.

Multiclass pattern recognition using adaptive correlation filters with complex constraints

An efficient method for reliable multiclass pattern recognition using a bank of adaptive correlation filters is proposed. The method can recognize and classify multiple targets from an input scene by using both the intensity and phase distributions of the output complex correlation plane. The adaptive filters are synthesized with the help of an iterative algorithm based on synthetic discriminant functions with complex constraints. The algorithm optimizes the discrimination capability of the adaptive filters and determines the minimum number of filters in a bank to guarantee a desired classification efficiency. As a result, the computational complexity of the proposed system is low. Computer simulation results obtained with the proposed approach in cluttered and noisy scenes are discussed and compared with those obtained through existing methods in terms of recognition performance, classification efficiency, and computational complexity.

Parallel support vector architectures for taxonomy of radial pulse morphology

Application of impedance plethysmography (IP) for impedance measurement is the paradigm in assessment of central and peripheral blood flow. We propose the expediency of IP to unearth hidden patterns from Plethysmographic observations on a radial pulse. The variability analysis in one thousand control and disease subjects evolves an archetype of eight different morphological patterns. The peripheral pulse waveforms not only characterize the physiology of control subjects, but also define the morphology of patients suffering from Myocardial Infarction, cirrhosis of liver, and disorder of lungs. Diverse parallel support vector machine (pSVM) topologies are designed as an aid to the physician for multiclass pattern recognition problem. Besides a lowest confusion coefficient (0.133), the PCA-based pSVM classifier offers a comparatively higher generalized correlation coefficient and κ value of 0.6586 and 0.8407, respectively. However, the ROC characteristics and the benchmark parameters suggest that wavelet-based pSVM is the optimum classifier with a sensitivity of 85.33 %, an elevated MCC (0.69), and a least upper bound on the expected error. pSVM stands out as a model classifier as compared with identified indices such as, Fishers Ratio, Morphology Index, and Heart rate variability.

Dissociable effects of methylphenidate, atomoxetine and placebo on regional cerebral blood flow in healthy volunteers at rest: A multi-class pattern recognition approach

The stimulant drug methylphenidate (MPH) and the non-stimulant drug atomoxetine (ATX) are both widely used for the treatment of attention deficit/hyperactivity disorder (ADHD), but their differential effects on human brain function are poorly understood. PET and blood oxygen level dependent (BOLD) fMRI have been used to study the effects of MPH and BOLD fMRI is beginning to be used to delineate the effects of MPH and ATX in the context of cognitive tasks. The BOLD signal is a proxy for neuronal activity and is dependent on three physiological parameters: regional cerebral blood flow (rCBF), cerebral metabolic rate of oxygen and cerebral blood volume. To identify areas sensitive to MPH and ATX and assist interpretation of BOLD studies in healthy volunteers and ADHD patients, it is therefore of interest to characterize the effects of these drugs on rCBF. In this study, we used arterial spin labeling (ASL) MRI to measure rCBF non-invasively in healthy volunteers after administration of MPH, ATX or placebo. We employed multi-class pattern recognition (PR) to discriminate the neuronal effects of the drugs, which accurately discriminated all drug conditions from one another and provided activity patterns that precisely localized discriminating brain regions. We showed common and differential effects in cortical and subcortical brain regions. The clearest differential effects were observed in four regions: (i) in the caudate body where MPH but not ATX increased rCBF, (ii) in the midbrain/substantia nigra and (iii) thalamus where MPH increased and ATX decreased rCBF plus (iv) a large region of cerebellar cortex where ATX increased rCBF relative to MPH. Our results demonstrate that combining ASL and PR yields a sensitive method for detecting the effects of these drugs and provides insights into the regional distribution of brain networks potentially modulated by these compounds.

Optimized code matrix generation for classification of multi-class pattern recognition problems using machine learning techniques

The pattern recognition applications like speech recognition, text classification and image recognition result in the solution of multi-class problems. Multi-class problems are reduced into several two class problems using the Machine Learning techniques such as Neural Networks and Support Vector Machines. We propose a hybrid approach for the design of output codes for multi-class pattern recognition problems. This approach has the advantage of taking into account the different aspects that are relevant for a code matrix to achieve good performance. Conventionally, code matrix is designed based on either the features of the problem or the features of the code matrix. The proposed work, focused on designing a new code matrix based on both the features of the problem and code matrix. This model aims at developing a hybrid version of ECOC and adaptive Recursive ECOC with BBO to achieve maximum classification accuracy and minimum computational time. Validation of the results has been performed using non-parametric statistical tests. The statistical results demonstrate that the evolving output codes through BBO provide a general-purpose method for improving the performance of base learners for real world multi-class pattern recognition problems.

A New Weighted Fuzzy C-Means Clustering Algorithm for Remotely Sensed Image Classification

Fuzzy clustering model is an essential tool to find the proper cluster structure of given data sets in pattern and image classification. In this paper, a new weighted fuzzy C-Means (NW-FCM) algorithm is proposed to improve the performance of both FCM and FWCM models for high-dimensional multiclass pattern recognition problems. The methodology used in NW-FCM is the concept of weighted mean from the nonparametric weighted feature extraction (NWFE) and cluster mean from discriminant analysis feature extraction (DAFE). These two concepts are combined in NW-FCM for unsupervised clustering. The main features of NW-FCM, when compared to FCM, are the inclusion of the weighted mean to increase the accuracy, and, when compared to FWCM, the centroid of each cluster is included to increase the stability. The motivation of this work is to meliorate the well-known fuzzy C-Means algorithm (FCM) and a recently proposed fuzzy weighted C-Means algorithm (FWCM). Our finding is that the proposed algorithm gives greater classification accuracy and stability than that of FCM and FWCM. Experimental results on both synthetic and real data demonstrate that the proposed clustering algorithm will generate better clustering results than those of FCM and FWCM algorithms, in particularly for hyperspectral images.

Evaluation of multiclass support vector machine classifiers using optimum threshold-based pruning technique

Support vector machine (SVM) is the state-of-the-art classifier used in real world pattern recognition applications. One of the design objectives of SVM classifiers using non-linear kernels is reducing the number of support vectors without compromising the classification accuracy. To meet this objective, decision-tree approach and pruning techniques are proposed in the literature. In this study, optimum threshold (OT)-based pruning technique is applied to different decision-tree-based SVM classifiers and their performances are compared. In order to assess the performance, SVM-based isolated digit recognition system is implemented. The performances are evaluated by conducting various experiments using speaker-dependent and multispeaker-dependent TI46 database of isolated digits. Based on this study, it is found that the application of OT technique reduces the minimum time required for recognition by a factor of 1.54 and 1.31, respectively, for speaker-dependent and multispeaker-dependent cases. The proposed approach is also applicable for other SVM-based multiclass pattern recognition systems such as target recognition, fingerprint classification, character recognition and face recognition.

Multinomial logistic regression and product unit neural network models: Application of a new hybrid methodology for solving a classification problem in the livestock sector

This work presents a new approach for multi-class pattern recognition based on the hybridization of a linear and nonlinear model. We propose multinomial logistic regression where some new covariates are defined by a product unit neural network, where in turn, the nonlinear basis functions are constructed with the product of the inputs raised to arbitrary powers. The application of this methodology involves, first of all, training the coefficients and the basis structure of product unit models using techniques based on artificial neural networks and evolutionary algorithms, followed by the application of multinomial logistic regression to both the new derived features and the original ones. To evaluate the efficacy of our technique we pose a difficult problem, the classification of sheep with respect to their milk production in different lactations, using covariates that only involve the first weeks of lactation. This enables the productive capacity of the animal to be identified more rapidly and leads to a faster selection process in determining the best producers. The results obtained with our approach are compared to other classification methodologies. Although several of these methodologies offer good results, the percentage of cases correctly classified was higher with our approach, which shows how instrumental the potential use of this methodology is for decision making in livestock enterprises, a sector relatively untouched by the technological innovations in business management that have been appearing in the last few years.

Pattern recognition with ordered labels

We consider a multi-class pattern recognition problem with linearly ordered labels and a loss function, which is a sum of squared deviations of decisions from true classes. It is proved that the optimal, in the Bayesian sense, decision rule is the a posteriori mean, which is then rounded to the nearest integer. Then, the plug-in type estimator of the optimal decision rule is proposed. We also investigate its asymptotic properties. Finally, we discuss briefly the structure of a neural net, which approximates the optimal decision rule.