Mahalanobis-Taguchi Research Articles

MT Method for Anomaly Detection and Classification using EM-λ Algorithm

Purpose: In this paper, we propose a method to classify and detect normal, known anomalies, and unknown anomalies by combining the expectation–maximisation (EM-λ) algorithm and the Mahalanobis–Taguchi (MT) method. Methodology/Approach: The proposed method learns normal data that are expected to be homogeneous and known abnormal data and performs classification and detection by parameter estimation using the EM-λ algorithm. Conventional methods perform analysis based on parameter estimation using the EM algorithm. However, the EM algorithm can degrade classification accuracy if it does not assume that the data fits the model's generative process. Findings: We verify the performance of the proposed method using artificially generated data and real-world bean data for classification as data that do not satisfy this assumption. The validation results show up to 6% improvement over the conventional method in classification accuracy and unknown anomaly discrimination accuracy. Research Limitation/implication: We try various patterns for the parameter of the proposed method in the verification. However, this way is computationally expensive. Originality/Value of paper: Conventional methods perform analysis based on parameter estimation using the EM algorithm. Our proposal method seeks to improve accuracy by using the EM-λ algorithm for parameter estimation, which is expected to improve classification accuracy when the data do not conform to the generative assumptions of the EM algorithm's model.

Multiclass Model for Agriculture Development Using Multivariate Statistical Method

Mahalanobis taguchi system (MTS) is a multi-variate statistical method extensively used for feature selection and binary classification problems. The calculation of orthogonal array and signal-to-noise ratio in MTS makes the algorithm complicated when more number of factors are involved in the classification problem. Also the decision is based on the accuracy of normal and abnormal observations of the dataset. In this paper, a multiclass model using Improved Mahalanobis Taguchi System (IMTS) is proposed based on normal observations and Mahalanobis distance for agriculture development. Twenty-six input factors relevant to crop cultivation have been identified and clustered into six main factors for the development of the model. The multiclass model is developed with the consideration of the relative importance of the factors. An objective function is defined for the classification of three crops, namely paddy, sugarcane and groundnut. The classification results are verified against the results obtained from the agriculture experts working in the field. The proposed classifier provides 100% accuracy, recall, precision and 0% error rate when compared with other traditional classifier models.

Anomaly detection for unlabelled unit space using the Mahalanobis–Taguchi system

Alongside the progress in technology related to the Internet of Things, the Mahalanobis–Taguchi (MT) system, which is an anomaly detection technique suitable for monitoring the condition of production equipment, has attracted attention. However, with the conventional MT method, historical data acquired and accumulated from sensors and smart devices cannot be analysed appropriately. This is because very often the accumulated historical information is data not labelled as either ‘normal’ or ‘anomaly’. Therefore, in this research, we propose a procedure that enables to detect anomalies with the MT method even when learning data are categorised as ‘unlabelled’. Specifically, we introduce a process for estimating the Mahalanobis distance on the population by applying a robust method based on γ divergence in the MT method. Through numerical experiments, we show that the proposed procedure is a useful anomaly detection technique for unlabelled data. Since this implies that labelling is redundant in anomaly detection, we conclude that the practicality of the MT method can be improved.

Anomaly detection in high-dimensional data with the Mahalanobis–Taguchi system

The Mahalanobis–Taguchi (MT) system is a typical Taguchi method and plays an important role in several fields. This study aims at improving the statistical procedure employed for anomaly detection in high-dimensional data with the MT system. The proposed study focuses on estimating the eigenvalues and eigenvectors of the covariance matrix and introduces an estimation procedure based on sparse principal component analysis (SPCA) in the MT system. By incorporating SPCA, eigenvalues and eigenvectors can be accurately estimated for high-dimensional data. In addition, the interpretation of the principal components can become simplified with decreasing number of nonzero elements in the estimated eigenvectors. Numerical experiments have confirmed that the proposed procedure is beneficial for both anomaly detection performance and investigating the cause of anomalies in high-dimensional data. Furthermore, a limitation of the proposed study is its emphasis on improving anomaly detection procedures founded on the first principal component and its residual component. However, the scope of such an anomaly detection procedure can be easily expanded for further improvement.

Two-step variable screening method for the Mahalanobis-Taguchi method with small training data

This paper proposes a method that applies the Mahalanobis-Taguchi (MT) method in situations where the number of variables exceeds the number of training samples. In the proposed method, the diagnostic effectiveness of the variables is first evaluated using signal-to-noise (SN) ratio, and the MT method is then applied using only those variables with high diagnostic effectiveness. Accordingly, the variables are selected via a two-step process: (1) variables are selected before creating the unit space and (2) conventional variable selection is carried out using the MT method. The results of experimental evaluation of the effectiveness of the proposed method using four different datasets from the UCI Machine Learning Repository indicate that its diagnostic performance is superior to that of the RT method.

A Simple Variable Screening Method for the Mahalanobis Taguchi Method

In this study, we describe a new variable selection method for the Mahalanobis Taguchi (MT) method. Typically, an orthogonal array is used for variable selection in the MT method. However, in this study, we developed a more intuitive and simpler variable selection method using a screening approach for identification of appropriate variables for diagnosis prior to creating the Mahalanobis space, in contrast to the conventional method, in which selection of the variables was carried out by repeated updating of the Mahalanobis space. Thus, the calculation in the proposed method was simpler than that in the conventional approach. We confirmed the effectiveness of the proposed method using a two-class diagnosis method (normal and abnormal). The results suggested that the proposed method had better diagnostic accuracy and robustness for changing supervised data.

Daily Peak Load Forecasting by Taguchi's T Method

SUMMARYIn this paper, a new technique of daily peak load forecasting is proposed. At first, we propose nonlinear correction T method as improvement technique of T method that is one of the multivariate analysis techniques, called Mahalanobis–Taguchi (MT) system, suggested in Quality Control. Moreover, we examine the daily peak load forecasting technique, which is based on data in the last year on a forecasting day, by T method and nonlinear correction T method. Furthermore, we examine the technique, which is based on the data of most recent several weeks of a forecasting day, by the T method and Nonlinear correction T method. In addition, the effectiveness of the proposed method by comparing the multiple regression analysis is discussed.

開発と研究 医療のInformation Technology化とMahalanobis-Taguchi法を用いたEvidence-Based Medicineの実践

開発と研究 医療のInformation Technology化とMahalanobis-Taguchi法を用いたEvidence-Based Medicineの実践

Estimation of Wire Bonding State by the Ensemble Based MT Method and Thin AE Sensor

With non-destructive, simple, and low-priced testing methods (which can be applied to manufacturing processes), we can assure high quality production by applying inspection across all products. Quality assurance in wire bonding is possible by using state diagnosis which utilizes detection of elastic waves close to the bonded joint. We succeeded in detecting elastic waves under difficult conditions and in harsh environments by using a thin AE (Acoustic Emission) sensor developed using AlN piezoelectric thin film technology. We attempted to conduct a wire bonding state estimation using the MT (Mahalanobis-Taguchi) method which is a simple and powerful pattern recognition technique. Two issues are raised when applying this method to the manufacturing process: 1) a large probability bias caused by an insufficient number of data samples, and 2) securing homogeneity of the Unit Space. In this paper, we present a state diagnosis method utilizing a real-world application which uses piezoelectric thin film sensing and an improved MT method to which ensemble learning is applied. Performance of this proposed method is also examined through a common benchmark dataset to assure efficient operation in any manufacturing process.

About the Evaluation of Liver Disease by the Monitoring of Mahalanobis Distance: Examination for Acute Hepatic Failure

It is Mahalanobis-Taguchi (MT) system to give the standard to evaluate a process of diagnosis of the doctor commonly. When Mahalanobis distance (MD) is calculated using MT system, it can be grasped as a unitary statistic. MD was calculated using data of fulminant hepatitis and acute on chronic hepatic failure, and a change of MD was compared with the outcome of patients. As a result, it was expressed for the numerical unified value that the data which changed complicatedly. When MD increase, the condition of a patient turns worse. When MD decreases, the condition of a patient is improved. This judgment was enabled easily and was useful as information in the medical treatment.

Type Classification Based on Inpatient Ratio According to Major Diagnosis Category of Group III in the Diagnosis Procedure Combination System.

The target was 1,283 hospitals in group III of medical institutions. Using the Diagnosis Procedure Combination (DPC) data and the inpatient ratio according to the major diagnosis category (MDC), group III hospitals were clarified on the basis of the Mahalanobis distance (D(2)) calculated using the Mahalanobis·Taguchi (MT) method. About 50 hospitals with the top inpatient ratio in each MDC and 1,233 other hospitals. The following were examined. 1) Significant difference between DPC data and inpatient ratio. 2) Distribution of D(2). 3) Classification based on the distribution of D(2). The top hospital was small scale, and there were few cases of emergency hospitalization, operation, general anesthesia induction, and chemotherapy. The number of cases involving the respiratory system, circulatory system, digestive and hepatobiliary systems, metabolic system, muscloskeletal system, kidney and urinary tracts was high with an inpatient ratio of more than 90%. On the bases of the distribution of D(2), the hospitals were classified into four types. Group one consisted of 213 special hospitals that showed high inpatient ratios of cases involving the nervous system, respiratory system, digestive and hepatobiliary systems, muscloskeletal system, injuries and burns. Group two consisted of 195 associate special hospitals that showed high inpatient ratios of cases involving the circulatory organ, metabolic system, kidney and urinary tracts, pediatric diseases, and mental diseases. Group three consisted of 223 hospitals for the seven remaining diseases. Group four consisted of 652 hospitals showing other distributions of D(2). Classification of hospitals to four types was possible by the MT method depending on the medical treatment results.

Examination of diagnosis procedure combination survey data that influence function evaluation coefficient II

On the basis of Diagnosis Procedure Combination (DPC) survey data, the factors that increase the value of function evaluation coefficient II were considered. A total of 1,505 hospitals were divided into groups I, II, and III, and the following items were considered. 1. Significant differences in function evaluation coefficient II and DPC survey data. 2. Examination of using the Mahalanobis-Taguchi (MT) method. 3. Correlation between function evaluation coefficient II and each DPC survey data item. 1. Function evaluation coefficient II was highest in group II. Group I hospitals showed the highest bed capacity, and numbers of hospitalization days, operations, chemotherapies, radiotherapies and general anesthesia procedures. 2. Using the MT method, we found that the number of ambulance conveyances was effective factor in group I hospitals, the number of general anesthesia procedures was effective factor in group II hospitals, and the bed capacity was effective factor in group III hospitals. 3. In group I hospitals, function evaluation coefficient II significantly correlated to the numbers of ambulance conveyances and chemotherapies. In group II hospitals, function evaluation coefficient II significantly correlated to bed capacity, the numbers of ambulance conveyances, hospitalization days, operations, general anesthesia procedures, and mean hospitalization days. In group III hospitals, function evaluation coefficient II significantly correlated to all items. The factors that improve the value of function evaluation coefficient II were the increases in the numbers of ambulance conveyances, chemotherapies and radiotherapies in group I hospitals, increases in the numbers of hospitalization days, operations, ambulance conveyances and general anesthesia procedures in group II hospitals, and increases in the numbers of hospitalization days, operations and ambulance conveyances. These results indicate that the profit of a hospital will increase, which will lead to medical services of good quality.

Exploiting the temporal patterning of transient VEP signals: a statistical single-trial methodology with implications to brain-computer interfaces (BCIs).

When visual evoked potentials (VEPs) are deployed in brain-computer interfaces (BCIs), the emphasis is put on stimulus design. In the case of transient VEPs (TVEPs) brain responses are never treated individually, i.e. on a single-trial (ST) basis, due to their poor signal quality. Therefore their main characteristic, which is the emergence during early latencies, remains unexplored. Following a pattern-analytic methodology, we investigated the possibility of using single-trial TVEP responses to differentiate between the different spatial locations where a particular visual stimulus appeared and decide whether it was attended or unattended by the subject. Covert spatial attention modulates the temporal patterning of TVEPs in such a way that a brief ST-segment, from a single synthesized sensor, is sufficient for a Mahalanobis-Taguchi (MT) system to decode subject's intention. In contrast to previous VEP-based approaches, stimulus-related information and user's intention are being decoded from transient ST-signals via exploiting aspects of brain response in the temporal domain. We demonstrated that in the TVEP signals there is sufficient discriminative information, coming in the form of a temporal code. We were able to introduce an efficient scheme that can fully exploit this information for the benefit of online classification. The measured performance brings high expectations for incorporating these ideas in BCI-control.

Authentication Based on Seating Pressure Distribution using the MT System

This paper proposes a system for authentication based on seating pressure distribution, using the MT system as a new method of biometric authentication that is difficult to forge and does not inconvenience users. The main characteristic is that the only action required of the user is to sit down. Feature values were extracted based on the pressure distribution when individuals seat, and individual users were distinguished from other persons by means of the Mahalanobis-Taguchi (MT) system used in quality engineering. The result of the experiment was a False Rejection Rate of 2.2% and a False Acceptance Rate of 1.1%.

Incorporating resting state dynamics in the analysis of encephalographic responses by means of the Mahalanobis–Taguchi strategy

The analysis of encephalographic responses has mostly been attempted via signal analytic techniques aiming at revealing the useful information from recordings which are considered as contaminated by the ubiquitous ongoing (or background) brain activity. There is continuously accumulating evidence for the existence of well-defined resting-state-networks (RSNs) in the brain, which play a crucial role in the generation of spontaneous activity and the associated neural responses. Hence, the signal plus noise is no longer a valid model and the ongoing fluctuations may influence the response.We introduce here the use of a multivariate statistical methodology, known as Mahalanobis–Taguchi (MT) strategy, which can be tailored to the spontaneous fluctuations so as to optimize the subsequent response detection. A subject-specific version of the MT strategy that combines the original methodology with a clustering algorithm for refining the training set is presented. The proposed methodology serves as an explorative tool for the detailed study of temporal patterning in brain responses.We demonstrate the potential of approach by applying it to experimental magneto-encephalographic (MEG) data. The results indicate vividly the effectiveness of the MT-strategy in analyzing and enhancing auditory responses.

The mahalanobis taguchi system—adaptive resonance theory neural network algorithm for dynamic product designs

This work presents a novel algorithm, which combines the Mahalanobis Taguchi system (MTS) with the adaptive resonance theory neural network (ARTN) method for parameter selections in a dynamic product design system (DPDS). The utility of the algorithm is assessed in two dimensions: the MTS shows how individual product parameter dimensions are selected, and the ARTN links parameter selection decisions across two different timelines and can be used to focus on DPDS and to identify product architecture dimensions that are critical for a DPDS. The MTS can easily solve product parameter selection problems and has proved to be computationally efficient in previous studies. Additionally, the ARTN algorithm provides a simple and efficient means of constructing a DPDS, which is verified by the ARTN algorithm and is presented in this study. From the results of this study, we conclude that the MTS-ARTN algorithm can be applied successfully in a DPDS environment.

Applying the Mahalanobis-Taguchi strategy for software defect diagnosis

The Mahalanobis-Taguchi (MT) strategy combines mathematical and statistical concepts like Mahalanobis distance, Gram-Schmidt orthogonalization and experimental designs to support diagnosis and decision-making based on multivariate data. The primary purpose is to develop a scale to measure the degree of abnormality of cases, compared to "normal" or "healthy" cases, i.e. a continuous scale from a set of binary classified cases. An optimal subset of variables for measuring abnormality is then selected and rules for future diagnosis are defined based on them and the measurement scale. This maps well to problems in software defect prediction based on a multivariate set of software metrics and attributes. In this paper, the MT strategy combined with a cluster analysis technique for determining the most appropriate training set, is described and applied to well-known datasets in order to evaluate the fault-proneness of software modules. The measurement scale resulting from the MT strategy is evaluated using ROC curves and shows that it is a promising technique for software defect diagnosis. It compares favorably to previously evaluated methods on a number of publically available data sets. The special characteristic of the MT strategy that it quantifies the level of abnormality can also stimulate and inform discussions with engineers and managers in different defect prediction situations.

Evaluation of the degree of deterioration of circuit‐breaker insulators using chemical analysis and the Mahalanobis–Taguchi (MT) method

AbstractWe have developed a novel evaluation technique of the deterioration degree of insulators for breakers using a chemical analysis and the Mahalanobis–Taguchi (MT) method. It is possible to evaluate the deterioration degree with great accuracy and nondestructively on‐site by this technique without the effect of humidity and external noise such as electromagnetic waves. The mechanism of the insulator's surface resistivity reduction was clarified, and it was found that the deterioration degree of insulators could be evaluated by this technique because a linear relationship existed between the results judged by the MT method and the actual measurement results for the surface resistivity. In comparison to electrical methods, such as partial electric discharge and megohmmeters, the evaluation accuracy has been improved by three digits and the range of the deterioration degree that could be evaluated was expanded by seven digits. © 2009 Wiley Periodicals, Inc. Electr Eng Jpn, 168(1): 11–20, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.20849

Remaining service life diagnostic technology of insulators for power distribution equipment

AbstractThe authors have developed a remaining service life diagnostic technology for insulators, which are important parts involved in determining the service life of power distribution equipment. Application of chemical evaluation and the Mahalanobis–Taguchi (MT) method enabled them to diagnose the insulators' degree of deterioration with great accuracy and nondestructively onsite. The theory derives the threshold value of surface resistivity for electric discharge initiation calculated based on the diagnostic results from the MT method and the year in which the master curve of service life and the threshold value intersect. It was clarified that a remaining service life could be predicted by this technology, because there was a good relationship between the check examination results of electric discharge initiation in the laboratory and the results the authors found. © 2009 Wiley Periodicals, Inc. Electr Eng Jpn, 167(2): 1–9, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.20854

Diagnosis of quality of fresh water for carbon steel corrosion by Mahalanobis distance

Corrosion diagnosis of carbon steel in fresh water was carried out by electrochemical method and Mahalanobis-Taguchi (MT) method. The corrosion morphology of carbon steel was classified by the combination of i– E curve measurement and chronopotentiometry. The MT method is one of discriminant analysis and can diagnose distinction data by multiple causes, i.e., ion concentrations. In the present paper, the concept for the application of this method to corrosion diagnosis is introduced. Additionally, electrochemical method was used for specific classification of carbon steel corrosion. By comparing the results of electrochemical measurements with those of Mahalanobis distance, the utility of MT method in the corrosion diagnosis was discussed.