Search

Implementing hardware accelerators of multiple classifier systems assures an improving in performance: on one hand, the combination of multiple classifiers outcomes is able to improve classification accuracy, with respect to a single classifier; on the other hand, implementing the prediction algorithm by means of an integrated circuit enables classifier systems with higher throughput and better latency compared with a pure software architecture. Although, this approach requires a very high amount of hardware resources, limiting the adoption of commercial configurable devices, such as Field Programmable Gate Arrays. In this paper, we exploit the application of Approximate Computing to trade classification accuracy off for hardware resources occupation. Specifically, we adopt the bit-width reduction technique on a multiple classifier system based on the Random Forest approach. A case study demonstrates the feasibility of the methodology, showing an area reduction ranging between 8.3 and 72.3%.

In this paper, we propose a novel approach for learning multi-label classifiers with the help of privileged information. Specifically, we use similarity constraints to capture the relationship between available information and privileged information, and use ranking constraints to capture the dependencies among multiple labels. By integrating similarity constraints and ranking constraints into the learning process of classifiers, the privileged information and the dependencies among multiple labels are exploited to construct better classifiers during training. A maximum margin classifier is adopted, and an efficient learning algorithm of the proposed method is also developed. We evaluate the proposed method on two applications: multiple object recognition from images with the help of implicit information about object importance conveyed by the list of manually annotated image tags; and multiple facial action unit detection from low-resolution images augmented by high-resolution images. Experimental results demonstrate that the proposed method can effectively take full advantage of privileged information and dependencies among multiple labels for better object recognition and better facial action unit detection.

In this paper we investigate the influence of outliers in the training set on the probabilistic classifier quality. By the example of naive Bayes classifier we show how the qualitative characteristics depend on the percentage of outliers' ratio. This dependence is built on three basic metrics of the classifier quality: precision, recall and F1 score. At the end we propose method for reducing the outliers influence on the classifier quality by approximating a piecewise linear function, and further using of gradient methods.

In our work we address limitations in the state-of-the-art in idiom type identification. We investigate different approaches for a lexical fixedness metric, a component of the state-of the-art model. We also show that our Machine Learning based approach to the idiom type identification task achieves an F1-score of 0.85, an improvement of 11 points over the state-of the-art.

In many problems of supervised tensor learning, real world data such as face images or MRI scans are naturally represented as matrices, which are also called as second order tensors. Most existing classifiers based on tensor representation, such as support tensor machine and kernelized support tensor machine need to solve iteratively which occupy much time and may suffer from local minima. In this paper, we present a kernel support matrix machine which performs a matrix-form inner product with maximum margin classifier. Specifically, the matrix inner product is introduced to leverage the inherent structural information within matrix data. Further, matrix kernel functions are applied to detect the nonlinear relationships. We analyze a unifying optimization problem for which we propose an asymptotically convergent algorithm. Theoretical analysis for the generalization bounds is derived based on Rademacher complexity with respect to a probability distribution. We demonstrate the merits of the proposed method by exhaustive experiments on both simulation study and a number of real-word datasets from a variety of application domains.

In this paper a fault tolerant probabilistic kernel version with smoothing parameter of Minsky’s perceptron classifier for more than two classes is exhibited and a correctness proof is provided. Moreover it is shown that the resulting classifier approaches optimality. Due to the non-determinism of the algorithm the (approximately) optimal value of a smoothing parameter has to be determined experimentally. The resulting complexity nevertheless allows for an efficient implementation employing for example Java concurrent programming and suitable hardware. In addition a probabilistic interpretation using Bayes Theorem is provided.

As an advanced local and global learning machine, the existing maxi---min margin machine (M4) still has its heavy time-consuming weakness. Inspired from the fact that covariance matrix of a dataset can characterize its data orientation and compactness globally, a novel large margin classifier called the local and global classification machine with collaborative mechanism (C2M) is constructed to circumvent this weakness in this paper. This classifier divides the whole global data into two independent models, and the final decision boundary is obtained by collaboratively combining two hyperplanes learned from two independent models. The proposed classifier C2M can be individually solved as a quadratic programming problem. The total training time complexity is $$O(2N^3)$$O(2N3) which is faster than $$O(N^4)$$O(N4) of M4. C2M can be well defined with the clear geometrical interpretation and can also be justified from a theoretical perspective. As an additional contribution, it is shown that C2M can robustly leverage the global information from those datasets with overlapping class margins, while M4 does not use such global information. We also use the kernel trick and exploit C2M's kernelized version. Experiments on toy and real-world datasets demonstrate that compared with M4, C2M is a more time-saving local and global learning machine.

Classifier System (MCS) is designed by combining two or more classifiers. MCS helps in increasing the accuracy of classification compared to the performance of the individual classifier. In this paper, Multiple Classifier System is implemented for automatic speech recognition. The combined classifier takes the final decision on predicted class label using a class label fuser (also called as classifier fuser). The class label fuser uses the predicted class labels of the two classifiers i.e Hidden Markov Model (HMM) and Support Vector Machines (SVM) and also involves the Dynamic Time Warping (DTW) technique for the final decision on the predicted label. There is an improvement in the accuracy of such classifier compared to the output of any individual classifier.

Twin support vector machines are a recently proposed learning method for pattern classification. They learn two hyperplanes rather than one as in usual support vector machines and often bring performance improvements. Semi-supervised learning has attracted great attention in machine learning in the last decade. Laplacian support vector machines and Laplacian twin support vector machines have been proposed in the semi-supervised learning framework. In this paper, inspired by the recent success of multi-view learning we propose multi-view Laplacian twin support vector machines, whose dual optimization problems are quadratic programming problems. We further extend them to kernel multi-view Laplacian twin support vector machines. Experimental results demonstrate that our proposed methods are effective.

Support vector machine is a beginning of the 1990s, based on statistical learning theory proposed new machine learning method, which structural risk minimization principle as the theoretical basis, by appropriately selecting a subset of functions and discriminant function in the subset, so the actual risk of learning machine to a minimum, to ensure that the limited training samples obtained through a small error classifier, an independent test set for testing error remains small. In this paper, support vector machine theory, algorithm, application status, etc. are discussed in detail.

Learning a classifier when only knowing the features and marginal distribution of class labels in each of the data groups is both theoretically interesting and practically useful. Specifically, we consider the case in which the ratio of the number of data instances to the number of classes is large. We prove sample complexity upper bound in this setting, which is inspired by an analysis of existing algorithms. We further formulate the problem in a density estimation framework to learn a generative classifier. We also develop a practical RBM-based algorithm which shows promising performance on benchmark datasets.

Purpose– The purpose of this paper is to understand the emotional state of a human being by capturing the speech utterances that are used during common conversation. Human beings except of thinking creatures are also sentimental and emotional organisms. There are six universal basic emotions plus a neutral emotion: happiness, surprise, fear, sadness, anger, disgust and neutral.Design/methodology/approach– It is proved that, given enough acoustic evidence, the emotional state of a person can be classified by an ensemble majority voting classifier. The proposed ensemble classifier is constructed over three base classifiers: k nearest neighbors, C4.5 and support vector machine (SVM) polynomial kernel.Findings– The proposed ensemble classifier achieves better performance than each base classifier. It is compared with two other ensemble classifiers: one-against-all (OAA) multiclass SVM with radial basis function kernels and OAA multiclass SVM with hybrid kernels. The proposed ensemble classifier achieves better performance than the other two ensemble classifiers.Originality/value– The current paper performs emotion classification with an ensemble majority voting classifier that combines three certain types of base classifiers which are of low computational complexity. The base classifiers stem from different theoretical background to avoid bias and redundancy. It gives to the proposed ensemble classifier the ability to be generalized in the emotion domain space.

Cascade classifiers are one of the most important contributions to real-time object detection. Nonetheless, there are many challenging problems arising in training cascade detectors. One common issue is that the node classifier is trained with a symmetric classifier. Having a low misclassification error rate does not guarantee an optimal node learning goal in cascade classifiers, i.e., an extremely high detection rate with a moderate false positive rate. In this work, we present a new approach to train an effective node classifier in a cascade detector. The algorithm is based on two key observations: 1) Redundant weak classifiers can be safely discarded; 2) The final detector should satisfy the asymmetric learning objective of the cascade architecture. To achieve this, we separate the classifier training into two steps: finding a pool of discriminative weak classifiers/features and training the final classifier by pruning weak classifiers which contribute little to the asymmetric learning criterion (asymmetric classifier construction). Our model reduction approach helps accelerate the learning time while achieving the pre-determined learning objective. Experimental results on both face and car data sets verify the effectiveness of the proposed algorithm. On the FDDB face data sets, our approach achieves the state-of-the-art performance, which demonstrates the advantage of our approach., Comment: 14 pages

In this paper we analyse the generalization performance of a type of binary classifier defined on the unit cube. This classifier combines some of the aspects of the standard methods that have been used in the logical analysis of data (LAD) and geometric classifiers, with a nearest-neighbor paradigm. We assess the predictive performance of the new classifier in learning from a sample, obtaining generalization error bounds that improve as a measure of `robustness' of the classifier on the training sample increases.

When a node of power system have a failure, we can deal with it by the multi-time electrical comprehensive phasor data of each node before and after the failure, introduces Bayesian Classifier to handle with the node failure case so as to distinguish and locate fault. Experiments show that Bayesian classifier is an effective method in fault location.

Accurate classification of text is a basic premise of extracting various types of information on the Web efficiently and utilizing the network resources properly. In this paper, a brand new text classification method was proposed. Consistency analysis method is a type of iterative algorithm, which mainly trains different classifiers (weak classifier) by aiming at the same training set, and then these classifiers will be gathered for testing the consistency degrees of various classification methods for the same text, thus to manifest the knowledge of each type of classifier. It main determines the weight of each sample according to the fact is the classification of each sample is accurate in each training set, as well as the accuracy of the last overall classification, and then sends the new data set whose weight has been modified to the subordinate classifier for training. In the end, the classifier gained in the training will be integrated as the final decision classifier. The classifier with consistency analysis can eliminate some unnecessary training data characteristics and place the key words on key training data. According to the experimental result, the average accuracy of this method is 91.0%, while the average recall rate is 88.1%.

An ensemble classifier system is a widely-used multi-classifier system, which combines the results from each classifier and, as a result, achieves better classification result than any single classifier used. Several methods have been used to build an ensemble classifier including boosting, which is a cascade method where misclassified examples in previous stage are used to boost the performance in current stage. Boosting is, however, a serial method which does not form a complete feedback loop. In this paper, proposed is context sensitive SVM ensemble (CASE) which adopts SVM, one of the best classifiers in term of classification rate, as a basic classifier and clustering method to divide feature space into contexts. As CASE divides feature space and trains SVMs simultaneously, the result from one component can be applied to the other and CASE achieves better result than boosting. Experimental results prove the usefulness of the proposed method.

The important requirements of designing a pattern classifier are high throughput and low memory requirements, and low cost hardware implementation. A pattern classifier by using Multiple Attractor Cellular Automata(MACA) proposed by Maji et al. reduced the complexity of the classification algorithm from to O(n) by using Dependency Vector(DV) and Dependency String(DS). In this paper, we generate a pattern classifier using LFSR to improve efficiently the space and time complexity and we propose a method for finding DV by using the 0-basic path. Also we investigate DV and the attractor of the generated pattern classifier. We can divide an n-bit DS by m number of s and generate various pattern classifiers.

The support vector machine (SVM), an assuring new method for the classification, has been widely used in many areas efficiently. However, the online learning issue of SVM is still not addressed satisfactorily since when a new sample arrives to retrain the SVM to adjust the classifier. This may not be feasible for real-time applications due to the expensive computation cost for re-training the SVM. This paper propose an Online SVM classifier algorithm known as OSVM-CH, which is based on the convex hull vertices selection depends on geometrical features of SVM. From the theoretical point of view, the first d+1(d is the dimension of the input samples) selected samples are proved to be vertices of the convex hull. This guarantees that the selected samples in our method keep the greatest amount of information of the convex hull. From the pedestrian detection application point of view, the new algorithm can update the classifier without reducing its classification performance.

We consider the classifier ensemble problem in this paper. Due to its superior performance to individual classifiers, class ensemble has been intensively studied in the literature. Generally speaking, there are two prevalent research directions on this, i.e., to diversely generate classifier components, and to sparsely combine multiple classifiers. While most current approaches are emphasized on either sparsity or diversity only, we investigate the classifier ensemble by learning both sparsity and diversity simultaneously. We manage to formulate the classifier ensemble problem with the sparsity or/and diversity learning in a general framework. In particular, the classifier ensemble with sparsity and diversity can be represented as a mathematical optimization problem. We then propose a heuristic algorithm, capable of obtaining ensemble classifiers with consideration of both sparsity and diversity. We exploit the genetic algorithm, and optimize sparsity and diversity for classifier selection and combination heuristically and iteratively. As one major contribution, we introduce the concept of the diversity contribution ability so as to select proper classifier components and evolve classifier weights eventually. Finally, we compare our proposed novel method with other conventional classifier ensemble methods such as Bagging, least squares combination, sparsity learning, and AdaBoost, extensively on UCI benchmark data sets and the Pascal Large Scale Learning Challenge 2008 webspam data. The experimental results confirm that our approach leads to better performance in many aspects.

an improved algorithm of Bayesian classifier combined with Fisher Linear Discriminant Analysis is proposed. This algorithm is the key to search the projection space of maximum separation. The original samples are projected to maximum separation space and new samples are obtained. These new samples are classifed by Bayes classifier. Experimental results show that improved Bayesian classifier has higher accuracy of classification and better performance of classification in the given data collection.

We suggest a new method called BRMM by modifying the existing RMM.The new method generalizes other well-known methods (SVM, RFDA, SVR).We prove and visualize the connections to the other methods.We suggest a sparse BRMM and prove an upper bound on the number of features. In this theoretical work we approach the class of relative margin classification algorithms from the mathematical programming perspective. In particular, we propose a Balanced Relative Margin Machine (BRMM) and then extend it by a 1-norm regularization. We show that this new classifier concept connects Support Vector Machines (SVM) with Fisher's Discriminant Analysis (FDA) by the insertion of a range parameter. It is also strongly connected to the Support Vector Regression. Using this BRMM it is now possible to optimize the classifier type instead of choosing it beforehand. We verify our findings empirically by means of simulated and benchmark data.

Water quality assessment is very significant to take timely measures for protecting water resources. This paper is to present a hybrid algorithm of fuzzy progressive transductive support vector classifier and bare-bones particle swarm optimization for water quality assessment. Then, we employ bare-bone particle swarm optimization to optimize the parameters of fuzzy progressive transductive support vector classifier, and the optimal parameters of fuzzy progressive transductive support vector classifier are obtained based on bare-bone particle swarm optimization. The experimental results indicate that the evaluation accuracy of bare-bones PSO-based fuzzy progressive transductive support vector classifier is higher than that of fuzzy progressive transductive support vector classifier, support vector classifier or artificial neural networks. Therefore, bare-bones PSO-based fuzzy progressive transductive support vector classifier is very suitable for water quality evaluation.

The importance of properly exploiting a classifier's inherent geometric characteristics when developing a classification methodology is emphasized as a prerequisite to achieving near optimal performance when carrying out thematic mapping. When used properly, it is argued that the long-standing maximum likelihood approach and the more recent support vector machine can perform comparably. Both contain the flexibility to segment the spectral domain in such a manner as to match inherent class separations in the data, as do most reasonable classifiers. The choice of which classifier to use in practice is determined largely by preference and related considerations, such as ease of training, multiclass capabilities, and classification cost.

In this paper, we propose a novel nonparallel hyperplane classifier, named ν-nonparallel support vector machine (ν-NPSVM), for binary classification. Based on our recently proposed method, i.e., nonparallel support vector machine (NPSVM), which has been proved superior to the twin support vector machines, ν-NPSVM is parameterized by the quantity ν to let ones effectively control the number of support vectors. By combining the ν-support vector classification and the ν-support vector regression together to construct the primal problems, ν-NPSVM inherits the advantages of ν-support vector machine so that enables us to eliminate one of the other free parameters of the NPSVM: the accuracy parameter e and the regularization constant C. We describe the algorithm, give some theoretical results concerning the meaning and the choice of ν, and also report the experimental results on lots of data sets to show the effectiveness of our method.

In this paper, we suggest classifier ensembles that can incorporate Support Vector Machine (SVM) as feature selection method into classifier ensembles models. Consequences of choosing different number of features are monitored. Also, the goal of this research is to present and compare different algorithmic approaches for constructing and evaluating systems that learn from experience to make the decisions and predictions and minimize the expected number or proportion of mistakes. Experimental results demonstrate the effectiveness of selecting features with SVM in various types of classifier ensembles.

In this paper, a region-partitioning active learning (AL) technique is proposed for classification of remote sensing (RS) images based on the support vector machines (SVM) classifier. In the batch-mode AL process, diversity information is required to select a batch of informative samples. A new AL technique that aims to introduce diversity information is proposed based on relative positions of candidate samples in the feature space. The proposed technique selects informative samples according to an uncertainty criterion at each iteration. These samples are selected with an extra constraint to guarantee that they are not located in the same region of the feature space. The proposed technique is compared with state-of-the-art methods adopted in the RS community. Experimental tests were performed on three data sets, including one very high spatial resolution multispectral data set and two hyperspectral data sets. The proposed algorithm displays a classification performance that is similar to or even better than the state-of-the-art methods. In addition, the proposed algorithm performs efficiently in terms of computational time.

A new privacy-preserving proximal support vector machine (P3SVM) is formulated for classification of vertically partitioned data. Our classifier is based on the concept of global random reduced kernel which is composed of local reduced kernels. Each of them is computed using local reduced matrix with Gaussian perturbation, which is privately generated by only one of the parties, and never made public. This formulation leads to an extremely simple and fast privacy-preserving algorithm, for generating a linear or nonlinear classifier that merely requires the solution of a single system of linear equations. Comprehensive experiments are conducted on multiple publicly available benchmark datasets to evaluate the performance of the proposed algorithms and the results indicate that: (a) Our P3SVM achieves better performance than the recently proposed privacy-preserving SVM via random kernels in terms of both classification accuracy and computational time. (b) A significant improvement of accuracy is attained by our P3SVM when compared to classifiers generated only using each party’s own data. (c) The generated classifier has comparable accuracy to an ordinary PSVM classifier trained on the entire dataset, without releasing any private data.

A current focus of intense research in pattern classification is the combination of several classifier systems, which can be built following either the same or different models and/or datasets building approaches. These systems perform information fusion of classification decisions at different levels overcoming limitations of traditional approaches based on single classifiers. This paper presents an up-to-date survey on multiple classifier system (MCS) from the point of view of Hybrid Intelligent Systems. The article discusses major issues, such as diversity and decision fusion methods, providing a vision of the spectrum of applications that are currently being developed.

In this article, the performance analysis of Expectation Maximization EM, Singular Value Decomposition SVD, and Support Vector Machines SVM classifiers for classification of carcinogenic regions from various medical images is carried out. Cancer detection is one of the critical issues where excessive care needs to be taken for better diagnosis. Any classifier needs to detect the cancer with respect to the efficiency in time of detection and performance. Due to these, three classifiers are selected: Expectation Maximization EM, Singular Value Decomposition SVD, and Support Vector Machines SVM. EM classifier performs as the optimizer and SVD classifier performs as the dual class classifier. SVM classifier is used as both optimizer and classifier for multiclass classification procedure and for wide stage cancer detection procedures. The performance analysis of all the three classifiers are analyzed for a group of 100 cancer patients based on the benchmark parameter such as Performance Measures and Quality Metrics. From the experimental results it is evident, that the SVM classifier significantly outperforms other classifiers in the classification of carcinogenic regions of medical images.

For real-time speech and audio encoders used in various multimedia applications, low-complexity encoding algorithms are required. Indeed, accurate classification of input signals is the key prerequisite for variable bit rate encoding, which has been introduced in order to effectively utilize limited communication bandwidth. This paper investigates implementation issues with a support vector machine (SVM)-based speech/music classifier in the selectable mode vocoder (SMV) framework, which is a standard codec adopted by the Third-Generation Partnership Project 2 (3GPP2). While a support vector machine is well known for its superior classification capability, it is accompanied by a high computational cost. In order to achieve a more realizable system, we propose two techniques for the SVM-based speech/music classifier, aimed at reducing the number of classification requests to the classifier. The first technique introduces a simpler classifier that processes some of the input frames instead of the SVM-based classifier, and the second technique skips a portion of input frames based on strong inter-frame correlation in speech and music frames. Our experimental results show that the proposed techniques can reduce the computational cost of the SVM-based classifier by 95.4 % with negligible performance degradation, making it plausible for integration into the SMV codec.

High accuracy and low overhead are two key features of a well-designed classifier for different classification scenarios. In this paper, we propose an improved classifier using a single-hidden layer feedforward neural network (SLFN) trained with extreme learning machine. The novel classifier first utilizes principal component analysis to reduce the feature dimension and then selects the optimal architecture of the SLFN based on a new localized generalization error model in the principal component space. Experimental and statistical results on the NSL-KDD data set demonstrate that the proposed classifier can achieve a significant performance improvement compared with previous classifiers.

The vector-matrix multiply and winner-take-all structure is presented as a general-purpose, low-power, compact, programmable classifier architecture that is capable of greater computation than a one-layer neural network, and equivalent to a two-layer perceptron. The classifier generates event outputs and is suitable for integration with event-driven systems. The main sources of mismatch, temperature dependence, and methods for compensation are discussed. We present measured data from simple linear and nonlinear classifier structures on a 0.35-μm chip and analyze the power and computing efficiency for scaled structures.

Mean Field Genetic Algorithm (MGA) is a hybrid algorithm of Mean Field Annealing (MFA) and Simulated annealing-like Genetic Algorithm (SGA). It combines benefit of rapid convergence property of MFA and effective genetic operations of SGA. This paper presents an approach for building a multi-classifier system in a MGA-based inductive learning environment. Multiple base classifiers are combined to build a multi-classifier system. A base classifier consists of a general classifier and a meta-classifier. The general classifier performs regular classification task. The meta-classifier evaluates classification result of its general classifier and decides whether the base classifier participates into a final decisionmaking process or not. The paper discusses our approach in details and presents some empirical results that show the improvement we can achieve with our approach.

The aims of the paper are multifold, to propose a new method to determine a suitable value of the bias corresponding to the soft margin SVM classifier and to experimentally evaluate the quality of the found value against one of the standard expression of the bias computed in terms of the support vectors. Also, it is proposed a variant of the Platt’s SMO algorithm to compute an approximation of the optimal solution of the SVM QP-problem. The new method for computing a more suitable value of the bias is based on genetic search. In order to evaluate the quality of the proposed method from the point of view of recognition and generalization rates, several tests were performed, some of the results being reported in the final section of the paper.

The Wisconsin Breast Cancer Dataset has been heavily cited as a benchmark dataset for classification. Neural Network techniques such as Neural Networks, Probabilistic Neural Networks, and Regression Neural Networks have been shown to perform very well on this dataset. However, despite its obvious practical importance and implications for cancer research, a thorough investigation of all modern classification techniques on this dataset remains to be done. In this paper we examine the efficacy of classifiers such as Random Forests with varying number of trees, Support Vector Machines with different kernels, Naive Bayes model and neural networks on the accuracy of classifying the masses in the dataset as benign/malignant. Results indicate that Support Vector machines with a Radial Basis function kernel give the best accuracy of all the models attempted. This indicates that there are non-linearities present in the dataset and that the Support vector machine does a good job of mapping the data into a higher dimensional space in which the non-linearities fade away and the data becomes linearly separable by large margin classifier like the support vector machine. These methods show that modern machine learning methods could provide for improved accuracy for early prediction of cancerous tumors.

Ensemble learning is to employ multiple individual classifiers and combine their predictions, which could achieve better performance than a single classifier. Considering that different base classifier gives different contribution to the final classification result, this paper assigns greater weights to the classifiers with better performance and proposes a weighted voting approach based on differential evolution. After optimizing the weights of the base classifiers by differential evolution, the proposed method combines the results of each classifier according to the weighted voting combination rule. Experimental results show that the proposed method not only improves the classification accuracy, but also has a strong generalization ability and universality.

The objective is to develop a dynamic decision selection method for face recognition system where minimum number of information about face are available to take correct decision. Statically we can develop such system where Bayesian method has been preferred in most case. It is better to fuse two or more classifier whose outputs are not highly correlated. In this work, the output of two classifiers systems are not so much correlated. We considered more than one decision for each classifier so that the correlations of the output are varied. It has been proved that the Bayesian optimal decision boundaries can be produced decision in fusion technique. It also has been proposed two methods to determine the Bayesian optimal decision are performed correctly in different database. We have proposed a different technique to calculate prior and posterior probability. Finally the fusion decision has been taken based on the probability values and it has been shown that the performance of Bayesian fusion techniques is better among the individual classifier technique. This fusion technique has been used in decision level and selected a class which is considered as correct output. Finally we have compared the performance among different classifier output and this soft-output classifier fusion method.

Classifier competence is critical important for dynamic classifier selection. This study proposes a semi-supervised learning algorithm to learn the competence of classifiers under the proposed optimization framework based on graph. First it constructs a graph based on the training data and some unlabeled data. Then it iteratively learns the competence of classifiers. The learned competence not just reflects the competitiveness of classifiers, but also varies smooth on the neighboring data. Experimental results on five different datasets show the dynamic classifier selection classification systems with the learned classifier competence perform better than the classification systems with local accuracy as the classifier competence.

Classification methods has become increasingly popular for biomedical and bioinformatical data analysis. However, due to the difficulty of data acquisition, sometimes we could only obtain small-scale datasets which may leads to unreasonable generalization performances. For SVM-like algorithms, we could resort to Large Margin theory to find out solutions for such dilemma. Recent studies on large margin theory show that, besides maximizing the minimum margin of a given training dataset, it is also necessary to optimization the margin distribution to boost the overall generalization ability. Correspondingly, a novel SVM-like algorithm called Large Margin Distribution Machine (LDM) realizes this idea by maximizing the average of margin and minimizing the variance of margin simultaneously. And a series of applications has been reported thereafter. There is another well-known machine learning algorithm called Extreme Learning Machine (ELM) which shares similar framework with SVM. It is believed in this paper ELM could also benefit from the virtues of margin distribution optimization. Bearing this in mind, a novel algorithm called Extreme Large Margin Distribution Machine(ELDM) is proposed in this paper by bridging the advantages of ELM and LDM. And an efficient extension of ELDM for multi-class classifications under One vs. All Scheme is proposed subsequently. Finally, the experiment results on both benchmark datasets and biomedical classification datasets show the effectiveness of our proposed algorithm.

The order-weighted averaging operator is commonly used in decision-making processes, where its powerful yet simple nature allows to aggregate output from several data sources into a meaningful result. Key to the order-weighted averaging operator is assignment of weights. Several approaches have been suggested and their properties examined for decision-making, except in the area of heuristic search assignment. In this paper, weight assignment is experimentally examined for supervised classification tasks using a fuzzy pattern classifier with order-weighted averaging, using maximum entropy and two separate heuristic search approaches, namely genetic algorithm and pattern search. The experiment is conducted using a sample of 20 UCI data sets. Results are discussed and recommendations made for when and how to apply heuristic search for this type of classifier.

The use of different evaluation measures for classification tasks have gained a significant amount of attention in the past decade, specially for those problems with multiple and imbalanced classes [1], [2]. However, the optimization of classifiers with respect to these measures is still heuristic, using ad-hoc rules with classical accuracy-optimized classifiers. We propose a classifier designed specifically to optimize one of the possible measures, namely, the so-called G-mean. Nevertheless, the technique is general, and it can be used to optimize generic evaluation measures. The optimization algorithm to train the classifier is described, and the numerical scheme is tested showing its usability and robustness. The code is publicly available, as well as the datasets used along this paper.

Machine learning plays a key role in many applications such as data mining and image recognition.Classification is one subcategory under machine learning. In this paper we propose a simple quantum circuitbased on the nearest mean classifier to classified handwriting characters. Our circuit is a simplified circuit fromthe quantum support vector machine [Phys. Rev. Lett. 114, 140504 (2015)] which uses quantum matrix inversealgorithm to find optimal hyperplane that separated two different classes. In our case the hyperplane is foundusing projections and rotations on the Bloch sphere.

Motor is a kind of imperative driving device, whether a motor can monitor its state precisely and diagnose fault timely have a profound impact. This paper mainly investigates the improvement of the general method of motor defect diagnosis to achieve higher accuracy. Unfortunately, every classifier has their own respective advantages and disadvantages, using the typical machine learning methods separately cannot achieve the expectant classify results. So, fusing the result of multiple classifiers to fully exploit the advantages of each sensor to reach the requirement of improving the classification accuracy. In this paper, three types of classifiers are fused: naive Bayes classifier, Random Forest classifier, and SVM classifier. By the algorithm of multi-classifier, the states of the motor can be predicted correctly.

Device-free localization (DFL) is expected to detect and locate a person by measuring the changes of received signals in wireless sensor networks without the need of any device. Fingerprint-based DFL in changeable environments has attracted wide attenuation in recent years. However, the accuracy of fingerprint-based localization could be improved further in changing environments. In this paper, we adopt the logistic regression classifier to counteract the bad influence to the localization in changeable environments by means of selecting the average channel. The experiment results show that the logistic regression classifier has a lower error rate than the k-nearest neighbors classifier and the linear discriminant analysis classifier. When the change of environments is very obvious, the logistic regression classifier achieves a better result than the random forests classifier in fingerprint-based localization.

Recognizing human activity is an important area of research in computer vision application. Manual monitoring of all cameras continuously for longer duration is inefficient making auto-detection of activity important. In this paper shape and optical flow features are fused together and used for human activity recognition. Features extracted are found to be efficient as concluded by ANOVA test. Hidden Markov Model are generated for each activity. System is trained and tested in various indoor and outdoor environment. The method adapted is made shape and angle invariant. Accuracy achieved using least square support vector machine classifier is 80% for all activities. Hidden Markov Model resulted in better accuracy as compared to least square support vector machine classifier with accuracy of 100.00% for walking, 100.00% for hand waving, 90% for bending, 84.61% for running and 90% for side gallop activities. 100% accuracy is achieved in recognizing activity in different angle with respect to camera.