Bayesian Classification Method Research Articles

Introduction of a methodology for visualization and graphical interpretation of Bayesian classification models.

Supervised machine learning models are widely used in chemoinformatics, especially for the prediction of new active compounds or targets of known actives. Bayesian classification methods are among the most popular machine learning approaches for the prediction of activity from chemical structure. Much work has focused on predicting structure-activity relationships (SARs) on the basis of experimental training data. By contrast, only a few efforts have thus far been made to rationalize the performance of Bayesian or other supervised machine learning models and better understand why they might succeed or fail. In this study, we introduce an intuitive approach for the visualization and graphical interpretation of naïve Bayesian classification models. Parameters derived during supervised learning are visualized and interactively analyzed to gain insights into model performance and identify features that determine predictions. The methodology is introduced in detail and applied to assess Bayesian modeling efforts and predictions on compound data sets of varying structural complexity. Different classification models and features determining their performance are characterized in detail. A prototypic implementation of the approach is provided.

Multivariate Analysis of Side Effects of Drug Molecules Based on Knowledge of Protein Bindings and ProteinProtein Interactions

Here, we examined the relationships between 969 side effects associated with 658 drugs and their 1368 human protein targets using our hybrid approaches. Firstly, L-shaped PLS (LPLS) was used to construct a multivariate model of side effects and protein bindings of drug molecules. LPLS is an extension of standard PLS regression, where, in addition to the response matrix Y and the regressor matrix X, an extra data matrix Z is constructed that summarizes the background information of X. X and Y are matrices comprising drugs-target proteins, and drugs-side effects, respectively. The Z matrix is the proteinprotein interaction data. From the loading plot of Y, we could identify two remarkable side effects (urinary incontinence and increased salivation) From the corresponding loading plot of X, the responsible protein targets causing each side effect could be estimated (sodium channels and gamma-aminobutyric acid (GABA) receptors). The loading plot of the Z matrix indicated that the GABA receptors interact with each other and they heavily influence the side effect of increased salivation. Secondly, Bayesian classifier methods were separately applied to the cases of the two side effects. That is, the Bayesian classifier method was used to classify drug molecules as binding or not binding to the responsible protein targets associated with each side effect. Using atom-coloring techniques, it was possible to estimate which fragments on the drug molecule might cause the specific side effects. This information is valuable for drug design to avoid specific side effects.

Performance Analysis using Bayesian Classification-Case Study of Software Industry

There has been rapid improvement in the ability to construct software systems , firstly by developing reliable hardware and second by developing effective process oriented methodologies. However there has been a lack of emphasis in people component of software engineering. This paper examines the human component and brings out interesting patterns for enhancing human performance. IT companies are constantly reaching out to find ways for software success and quality and innovations for having a competitive edge over other companies. It has been found that human aspects seem to dominate the most expensive project failures. Therefore Human Aspect of Software Engineering has emerged as a new dimension in software engineering in recent years. IT industry have a large set of data having precious information related to project personnel. Data mining technique has the ability to discover knowledge from this unexplored data. In this paper, data mining techniques particularly Bayesian classification method is applied on the data related to team members of a software company to help an organization to find the right project personnel who will contribute largely to project success by performing well. This study will help the organization to reduce the failure ratio of software to a significant level and enhance the quality of the software by deploying the right person at the very start of all software process.

Reserch on a NO<sub>x</sub> Emission Forecasting Method Based on Naïve Bayesian

Increasingly serious environmental pollution,trying to find a effective method to control NOx emission become more importance. Under this background, this paper adopts the naïve Bayesian classifier method which built on the basis of the probability density function to forecasting the NOx emission of diesel engine. This paper proposes a new approach to weight the super-parent one dependence estimators, and uses the UCI datasets to verify the validity of the proposed method. Finally, apply this diagnosis technology to the collected WD615 diesel engine data. The comparison experiments with other algorithms demonstrate the effectiveness of the proposed method.

Sparse Gaussian processes with manifold-preserving graph reduction

Gaussian processes are a popular and effective Bayesian method for classification and regression. Due to the cube time complexity with respect to the size of the training set, time consumption will be too high for Gaussian processes to deal with big data sets. Based on the manifold assumption, manifold-preserving graph reduction (MPGR) is a simple and efficient graph sparsification method originally proposed for sparse semi-supervised learning. High representative points are selected, while outliers and noisy points are excluded by MPGR with low time consumption. In this paper, we apply MPGR to sparse supervised learning, and utilize MPGR to get a sparse representation of Gaussian processes. A fatal shortcoming of MPGR is that it does not effectively consider the influence of outputs to Gaussian processes. We proceed to exploit outputs for sparse Gaussian processes by embedding a method that considers outputs in the MPGR framework, and propose manifold-preserving graph reduction with outputs (MPGRO) for sparse Gaussian processes. In this paper, we utilize informative vector machine (IVM) for MPGRO. Experimental results on seven real data sets show that MPGR is better than IVM for sparse Gaussian processes, and MPGRO is better than MPGR and IVM for sparse Gaussian processes and is comparable on error rates to the method utilizing full training sets.

Bayesian Classification of Hyperspectral Imagery Based on Probabilistic Sparse Representation and Markov Random Field

This letter presents a Bayesian method for hyperspectral image classification based on the sparse representation (SR) of spectral information and the Markov random field modeling of spatial information. We introduce a probabilistic SR approach to estimate the class conditional distribution, which proved to be a powerful feature extraction technique to be combined with the label prior distribution in a Bayesian framework. The resulting maximum a priori problem is estimated by a graph-cut-based α-expansion technique. The capabilities of the proposed method are proven in several benchmark hyperspectral images of both agricultural and urban areas.

Eigenvalue Analysis-Based Approach for POL-SAR Image Classification

A novel polarimetric synthetic aperture radar (POL-SAR) image classification approach is proposed in this paper by exploiting coherency matrix eigenvalues for polarimetric information representation and understanding. The approach consists of two parts. Initially, the statistical distributions of eigenvalue for homogeneous areas are analyzed by taking eigenvalues as the features of polarimetric information. The Bayesian classification method is applied to verify the feasibility of distinguishing different homogeneous areas. As a result, this method can work well those pixels with the similar scatter mechanism by using different polarimetric intensity information from eigenvalues. But this process cannot adequately distinguish those pixels with similar eigenvalues distribution. So, an eigenvalues-based local operator is defined to overcome the insufficient of the similar pixels by introducing a similar measure and eigenvalues-based texture information. After all pixels are classified by Bayesian classification, if the similarity of the pixel is larger than the given threshold, this pixel will be further classified by support vector machine using texture information. The proposed method is tested on three POL-SAR datasets, in which the average classification accuracy of eight categories for the Flevoland data from our method reaches nearly 90%.

Incremental Naïve Bayesian Learning Algorithm based on Classification Contribution Degree

In order to improve the ability of gradual learning on the training set gotten in batches of Naive Bayesian classifier, an incremental Naive Bayesian learning algorithm is improved with the research on the existing incremental Naive Bayesian learning algorithms. Aiming at the problems with the existing incremental amending sample selection strategy, the paper introduced the concept of sample Classification Contribution Degree in the process of incremental learning, based on the comprehensive consideration about classification discrimination, noisy and redundancy of the new training data. The definition and theoretical analysis of sample Classification Contribution Degree is given in this paper. Then the paper proposed the incremental Naive Bayesian classification method based on the Classification Contribution Degree. The experimental results show that the algorithm simplified the incremental learning process, improved the classification accuracy of incremental learning.

Adjacent Total Average Algorithm for Travel Time Prediction

Travel time prediction plays an important role with the development of ATIS (Advanced Travelers Information Systems).It provides useful information which may allow travelers to change their routes and to decide whether or not to make necessary changes to their routes or departure times. Here we proposed a new method ATA(Adjacent total average). The challenge of this paper is to provide accurate travel time to compare with others method .With the use of same set of historical travel time and comparing the previous results of other four methods like Successive Moving Average (SMA), Naive Bayesian Classification (NBC) method, Switching method and K -means Clustering(MKC) Our algorithms exhibit high accuracy in predicting travel time.

Recursive Actuator Fault Detection and Diagnosis for Emergency Landing of UASs

This paper presents a practical recursive fault detection and diagnosis (FDD) scheme for online identification of actuator faults for unmanned aerial systems (UASs) based on the unscented Kalman filtering (UKF) method. The proposed FDD algorithm aims to monitor health status of actuators and provide indication of actuator faults with reliability, offering necessary information for the design of fault-tolerant flight control systems to compensate for side-effects and improve fail-safe capability when actuator faults occur. The fault detection is conducted by designing separate UKFs to detect aileron and elevator faults using a nonlinear six degree-of-freedom (DOF) UAS model. The fault diagnosis is achieved by isolating true faults by using the Bayesian Classifier (BC) method together with a decision criterion to avoid false alarms. High-fidelity simulations with and without measurement noise are conducted with practical constraints considered for typical actuator fault scenarios, and the proposed FDD exhibits consistent effectiveness in identifying occurrence of actuator faults, verifying its suitability for integration into the design of fault-tolerant flight control systems for emergency landing of UASs.

Non-destructive evaluation of checking in thermally modified timber

Non-destructive measurement techniques were used to study the characteristics of thermally modified timber (TMT), especially checking. Two non-destructive measurement techniques were evaluated: air-coupled ultrasound (ACU) and electrical impedance spectroscopy (EIS). In the studied TMT (N = 38), density and pith location in the cross-section were positively correlated with the number of checks in the cross-section. Several ultrasound signal features correlated with TMT characteristics, most strongly with number of checks. Severely checked samples and other samples were considered as two different classes of checking, and the potential of ultrasound for differentiating between the two classes was tested by using Bayesian classification method. The correct classification rate was 97 %. EIS parameters correlated with density and latewood content of TMT more strongly than the ACU parameters. Thus, for determination of checking and density or earlywood/latewood content, both EIS and ACU measurements are recommended.

A New PLS and Bayesian Classification Based On-Line Outlier Detection Method

Considering that multivariate data is difficult to detect, this paper propose an PLS and Bayesian theory based on-line outlier detection method. Firstly, it figures out the Q-statistics by PLS(partial least squares analysis), then classify Q statistics with Bayesian classification method and decide whether or not the sample data is normal. We employ UCI database to verify the method, the simulation results show that, compared to traditional PCA based method, it has lower ratio of error judgement, and is more effective in detecting outliers and identifying the change of process states.

Combined intelligent methods based on wireless sensor networks for condition monitoring and fault diagnosis

This study presents new combined methods based on multiple wireless sensor system for real-time condition monitoring of electric machines. The established experimental setup measures multiple signals such as current and vibration on a common wireless node. The proposed methods are low-cost, intelligent, and non-intrusive. The proposed wireless network based framework is useful for analyzing and monitoring of signals from multiple induction motors. Motor current and vibration signals are simultaneously read from multiple motors through wireless nodes and the faults are estimated using two combined methods. Phase space analysis of vibration data and amplitudes of three phase current signals are used as features in combined intelligent classifiers. Stator related faults are diagnosed by analyzing the magnitudes of read current signals with fuzz logic. The vibration signal taken from the two-axis acceleration meter is normalized and phase space of this signal is constructed. The change in phase spaces are analyzed with machine learning techniques based on Gaussian Mixture Models and Bayesian classification to detect bearing faults. The phase space of vibration signals is constructed by using non-linear time series analysis and Gaussian mixtures are obtained for healthy and each faulty conditions. The constructed mixture models are classified according to their distribution on phase space by using Bayesian classification method. Four motor operating conditions- stator open phase fault, one and two bearing imbalance faults, and healthy condition are considered and related signals are obtained to evaluate the proposed system. The accuracy of the proposed system is confirmed by experimental data.

Computer-aided diagnosis system: A Bayesian hybrid classification method

A novel method to classify multi-class biomedical objects is presented. The method is based on a hybrid approach which combines pairwise comparison, Bayesian regression and the k-nearest neighbor technique. It can be applied in a fully automatic way or in a relevance feedback framework. In the latter case, the information obtained from both an expert and the automatic classification is iteratively used to improve the results until a certain accuracy level is achieved, then, the learning process is finished and new classifications can be automatically performed. The method has been applied in two biomedical contexts by following the same cross-validation schemes as in the original studies. The first one refers to cancer diagnosis, leading to an accuracy of 77.35% versus 66.37%, originally obtained. The second one considers the diagnosis of pathologies of the vertebral column. The original method achieves accuracies ranging from 76.5% to 96.7%, and from 82.3% to 97.1% in two different cross-validation schemes. Even with no supervision, the proposed method reaches 96.71% and 97.32% in these two cases. By using a supervised framework the achieved accuracy is 97.74%. Furthermore, all abnormal cases were correctly classified.

The Defect Identification of LED Chips Based on Bayesian Classifier

The types, causes and image properties of the defective and damaged chips are been analyzes. On the basis of analyzing the image properties, Bayesian classifier and identification method are proposed. Based on the Bayesian decision theory, five kinds of image features such as the number of light spot, dark spot, edge spot, block and area are applied. Normal distribution is made use to building classifier and studying the correction strategy of model. In the process of identification, dynamical correction and model parameters completion make the classifier have some degree of self-learning feature and higher identification accuracy. Keywords: defect; feature; classifier; decision

Breaking the curse of dimensionality in quadratic discriminant analysis models with a novel variant of a Bayes classifier enhances automated taxa identification of freshwater macroinvertebrates

Macroinvertebrate samples are commonly used in biomonitoring to study changes on aquatic ecosystems. Traditionally, specimens are identified manually to taxa by human experts being time‐consuming and cost intensive. Using the image data of 35 taxa and 64 features, we propose a novel variant of the quadratic discriminant analysis for breaking the curse of dimensionality in quadratic discriminant analysis models. Our variant, called a random Bayes array (RBA), uses bagging and random feature selection similar to random forest. We explore several variations of RBA. We consider three classification (i.e taxa identification) decisions: majority vote, averaged posterior probabilities, and a novel approach; a score of weighted votes. Besides modifying the voting, we propose to weight features according to their importance instead of eliminating the least important features. We compared the performance of RBA with traditional Bayesian and several other popular classification methods and assessed how the methods behave in relation to each other and the different macroinvertebrate species. Further, we investigate how severely misclassifications affect the performance of different methods when set into a biomonitoring context. We found that the lowest and least severe classification error (i.e. most accurate taxa identification) was achieved with RBA by using averaged posterior probabilities and weighted features. Copyright © 2013 John Wiley & Sons, Ltd.

A Bayesian Network Method for Automatic Classification of Eddy Current NDE Signals

Eddy current nondestructive evaluation (ECNDE) techniques are widely used in structural integrity and health monitoring. A novel algorithm was proposed for characterizing eddy current (EC) signals. In scanning inspection, the EC signals responding to impedance change were pre-processed for noise elimination and feature extraction. After feature extraction, Bayesian networks (BNs) were carried out to classify EC signals. It is shown by extensive experiments that kernel principal component analysis (KPCA) is better than principal component analysis (PCA) for feature extraction. The methods using BNs by KPCA feature extraction can perform better than the other classification methods.

Automatic Sputum Color Image Segmentation for Lung Cancer Diagnosis

Lung cancer is considered to be the leading cause of cancer death worldwide. A technique commonly used consists of analyzing sputum images for detecting lung cancer cells. However, the analysis of sputum is time consuming and requires highly trained personnel to avoid errors. The manual screening of sputum samples has to be improved by using image processing techniques. In this paper we present a Computer Aided Diagnosis (CAD) system for early detection and diagnosis of lung cancer based on the analysis of the sputum color image with the aim to attain a high accuracy rate and to reduce the time consumed to analyze such sputum samples. In order to form general diagnostic rules, we present a framework for segmentation and extraction of sputum cells in sputum images using respectively, a Bayesian classification method followed by region detection and feature extraction techniques to determine the shape of the nuclei inside the sputum cells. The final results will be used for a (CAD) system for early detection of lung cancer. We analyzed the performance of a Bayesian classification with respect to the color space representation and quantification. Our methods were validated via a series of experimentation conducted with a data set of 100 images. Our evaluation criteria were based on sensitivity, specificity and accuracy.

Texture attributes for detection of salt

Abstract Within areas of salt tectonics, seismic imaging requires extensive updating of the velocity model. This includes defining the boundaries of salt structures which are often characterized by changes in texture of the seismic signal rather than reflectivity. The main characteristics of texture inside salt structures are identified. Three groups of texture attributes are studied: gray-level co-occurrence matrix (GLCM) attributes, frequency-based attributes, and dip and similarity attributes. Various combinations of the selected attributes are tested in a supervised Bayesian classification method. Experimental results show that the classification performance improves by combining at least two texture attribute groups. The classifier computes an estimate of the pixelwise probability of salt. It can then be applied to compute the probability of salt on different seismic sections. Classification results were found more robust based on timeslices. The result from classification, the salt probability image, is then input to a segmentation algorithm that produces a smooth border delimiting the extent of the salt. The segmented salt contours corresponded fairly well to the contours provided by an interpreter.

Sloping farmland identification using hierarchical classification in the Xi-He region of China

The Loess Plateau suffers the most serious soil erosion in China. Sloping cultivated land is one of the most common land types in the region, and it leads to severe soil erosion. Analyses based on fine resolution satellite imagery can play a key role in the surveying of sloping farmland. In this article, a combination of remote-sensing (RS) and geographical information system (GIS) techniques under the hierarchical classification framework is used to investigate the sloping cultivated land in the Xi–He ecological engineering demonstration region of the Loess Plateau. This article synthetically adopts Système Pour l'Observation de la Terre 5 (SPOT-5) high-resolution images, Thematic Mapper (TM) multi-spectral images, and a terrain map at a scale of 1:10,000 to calculate the spectral, textural, and slope attributes of the sloping land type. A Bayesian classification method is employed to distinguish the crop and non-crop areas with a priori knowledge provided by local experts. The Gabor filter and the grey-level co-occurrence matrices (GLCM) method are used in combination under a man–machine interactive interpretation to extract the differences of texture trends and density distributions between sloping cultivated land and terraces, and a window-based method of texture analysis operations is adopted. A classification accuracy assessment by field survey shows that the total interpretation accuracy exceeds 80%. The sloping cultivated land in the research region covers an area of 233.06 km2, and the sub-watersheds, such as the Yangliu watershed, need key management in the future.