An unsupervised one-class-classifier support vector machine to simultaneously monitor location and scale of multivariate quality characteristics

PurposeThe purpose of this study is to propose an effective unsupervised one-class-classifier (OCC) support vector machine (SVM)-based single multivariate control chart (OCC-SVM) to simultaneously monitor “location” and “scale” shifts of a manufacturing process.Design/methodology/approachThe step-by-step approach to developing, implementing and fine-tuning the intrinsic parameters of the OCC-SVM chart is demonstrated based on simulation and two real-life case examples.FindingsA comparative study, considering varied known and unknown response distributions, indicates that the OCC-SVM is highly effective in detecting process shifts of samples with individual observations. OCC-SVM chart also shows promising results for samples with a rational subgroup of observations. In addition, the results also indicate that the performance of OCC-SVM is unaffected by the small reference sample size.Research limitations/implicationsThe sample responses are considered identically distributed with no significant multivariate autocorrelation between sample observations.Practical implicationsThe proposed easy-to-implement chart shows satisfactory performance to detect an out-of-control signal with known or unknown response distributions.Originality/valueVarious multivariate (e.g. parametric or nonparametric) control chart(s) are recommended to monitor the mean (e.g. location) and variance (e.g. scale) of multiple correlated responses in a manufacturing process. However, real-life implementation of a parametric control chart may be complex due to its restrictive response distribution assumptions. There is no evidence of work in the open literature that demonstrates the suitability of an unsupervised OCC-SVM chart to simultaneously monitor “location” and “scale” shifts of multivariate responses. Thus, a new efficient OCC-SVM single chart approach is proposed to address this gap to monitor a multivariate manufacturing process with unknown response distributions.

Cyclical Learning Rates (CLR’s) for Improving Training Accuracies and Lowering Computational Cost

Prediction of different lung pathologies using chest X-ray images is a challenging task requiring robust training and testing accuracies. In this article, one-class classifier (OCC) and binary classification algorithms have been tested to classify 14 different diseases (atelectasis, cardiomegaly, consolidation, effusion, edema, emphysema, fibrosis, hernia, infiltration, mass, nodule, pneumonia, pneumothorax and pleural-thickening). We have utilized 3 different neural network architectures (MobileNetV1, Alexnet, and DenseNet-121) with four different optimizers (SGD, Adam, and RMSProp) for comparing best possible accuracies. Cyclical learning rate (CLR), a tuning hyperparameters technique was found to have a faster convergence of the cost towards the minima of cost function. Here, we present a unique approach of utilizing previously trained binary classification models with a learning rate decay technique for re-training models using CLR’s. Doing so, we found significant improvement in training accuracies for each of the selected conditions. Thus, utilizing CLR’s in callback functions seems a promising strategy for image classification problems.

Feature Extraction of Sequence of Keystrokes in Fixed Text Using the Multivariate Hawkes Process

In this paper, we propose a new method of extracting the features of keystrokes. The Hawkes process based on exponential excitation kernel was used to model the sequence of keystrokes in fixed text, and the intensity function vector and adjacency matrix of the model obtained through training were regarded as the characteristics of the keystrokes. A visual analysis was carried out on the CMU keystroke raw data and the feature data extracted using the proposed method. We used one-class classifier to compare the classification effect of CMU keystroke raw data and the feature data extracted by the Hawkes process model and POHMM model. The experimental results show that the feature data extracted using the proposed method contains rich information to distinguish users. In addition, the feature data extracted using the proposed method has a slightly better classification performance than the original CMU keystroke data for some users who are not easy to distinguish.

Extrapolating Satellite-Based Flood Masks by One-Class Classification—A Test Case in Houston

Flood masks are among the most common remote sensing products, used for rapid crisis information and as input for hydraulic and impact models. Despite the high relevance of such products, vegetated and urban areas are still unreliably mapped and are sometimes even excluded from analysis. The information content of synthetic aperture radar (SAR) images is limited in these areas due to the side-looking imaging geometry of radar sensors and complex interactions of the microwave signal with trees and urban structures. Classification from SAR data can only be optimized to reduce false positives, but cannot avoid false negatives in areas that are essentially unobservable to the sensor, for example, due to radar shadows, layover, speckle and other effects. We therefore propose to treat satellite-based flood masks as intermediate products with true positives, and unlabeled cells instead of negatives. This corresponds to the input of a positive-unlabeled (PU) learning one-class classifier (OCC). Assuming that flood extent is at least partially explainable by topography, we present a novel procedure to estimate the true extent of the flood, given the initial mask, by using the satellite-based products as input to a PU OCC algorithm learned on topographic features. Additional rainfall data and distance to buildings had only minor effect on the models in our experiments. All three of the tested initial flood masks were considerably improved by the presented procedure, with obtainable increases in the overall κ score ranging from 0.2 for a high quality initial mask to 0.7 in the best case for a standard emergency response product. An assessment of κ for vegetated and urban areas separately shows that the performance in urban areas is still better when learning from a high quality initial mask.

A two-stage classification algorithm for radar targets based on compressive detection

Algorithms are proposed to address the radar target detection problem of compressed sensing (CS) under the conditions of a low signal-to-noise ratio (SNR) and a low signal-to-clutter ratio (SCR) echo signal. The algorithms include a two-stage classification for radar targets based on compressive detection (CD) without signal reconstruction and a support vector data description (SVDD) one-class classifier. First, we present the sparsity of the echo signal in the distance dimension to design a measurement matrix for CD of the echo signal. Constant false alarm rate (CFAR) detection is performed directly on the CD echo signal to complete the first-order target classification. In simulations, the detection performance is similar to that of the traditional matched filtering algorithm, but the data rate is lower, and the necessary data storage space is reduced. Then, the power spectrum features are extracted from the data after the first-order classification and converted to the feature domain. The SVDD one-class classifier is introduced to train and classify the characteristic signals to complete the separation of the targets and the false alarms. Finally, the performance of the algorithm is verified by simulation. The number of false alarms is reduced, and the detection probability of the targets is improved.

MultiKOC: Multi-One-Class Classifier Based K-Means Clustering

In the computational biology community there are many biological cases that are considered as multi-one-class classification problems. Examples include the classification of multiple tumor types, protein fold recognition and the molecular classification of multiple cancer types. In all of these cases the real world appropriately characterized negative cases or outliers are impractical to achieve and the positive cases might consist of different clusters, which in turn might lead to accuracy degradation. In this paper we present a novel algorithm named MultiKOC multi-one-class classifiers based K-means to deal with this problem. The main idea is to execute a clustering algorithm over the positive samples to capture the hidden subdata of the given positive data, and then building up a one-class classifier for every cluster member’s examples separately: in other word, train the OC classifier on each piece of subdata. For a given new sample, the generated classifiers are applied. If it is rejected by all of those classifiers, the given sample is considered as a negative sample, otherwise it is a positive sample. The results of MultiKOC are compared with the traditional one-class, multi-one-class, ensemble one-classes and two-class methods, yielding a significant improvement over the one-class and like the two-class performance.

A One-Class-Classifier-Based Negative Data Generation Method for Rapid Earthquake-Induced Landslide Susceptibility Mapping

Machine learning with extensively labeled training samples (e.g., positive and negative data) has received much attention in terms of addressing earthquake-induced landslide susceptibility mapping (LSM). However, the extensive amount of labeled training data required by machine learning, particularly the precise negative data (i.e., non-landslide area), cannot be easily and efficiently collected. To address this issue, this study presents a one-class-classifier-based negative data generation method for rapid earthquake-induced LSM. First, an incomplete landslide inventory (i.e., positive data) was produced with the aid of change detection using before-and-after satellite images and the Geographic Information System (GIS). Second, a one-class classifier was utilized to compute the probability of landslide occurrence based on the incomplete landslide inventory followed by the negative data generation from the low landslide susceptibility areas. Third, the positive data as well as the generated negative data (i.e., non-landslide) were compounded to train a traditional binary classifier to produce the final LSM. Experimental results suggest that the proposed method is capable of achieving a result that is comparable to methods using the complete landslide inventory, and it displays good correspondence with recent landslide events, making it a suitable method for rapid earthquake-induced LSM. The findings in this study would be useful in regional disaster planning and risk reduction.

Recognizing the Style of Artistic Painting via Information Entropy for Smart City Construction

Digitalization is conducive to the protection and inheritance of culture and civilization. The artistic painting recognition is an essential part in digitalization and plays an important role in smart city construction. This paper proposes a novel framework to recognize Chinese painting style by using information entropy. First, the authors choose the ink painting, pyrography, mural, and splash ink painting as the known artistic styles. Then, this article uses the information entropy to represent the paintings. The information entropy includes color entropy, block entropy, and contour entropy. The color entropy is obtained by a weighted function of Channel A and B in the lab color space. The block entropy is the average information entropy of blocks which are a small part of the image. The contour entropy is obtained from the contour information which is obtained by contourlet transform. The information entropy is input into an oracle to determine the style. The oracle includes a one-class classifier and a classical classifier. The effectiveness is verified on the real painting set.

A One-class Classifier Based on a Hybrid Topology to Detect Faults in Power Cells

The use of batteries became essential in our daily life in electronic devices, electric vehicles and energy storage systems in general terms. As they play a key role in many devices, their design and implementation must follow a thorough test process to check their features at different operating points. In this circumstance, the appearance of any kind of deviation from the expected operation must be detected. This research deals with real data registered during the testing phase of a lithium iron phosphate—LiFePO4—battery. The process is divided into four different working points, alternating charging, discharging and resting periods. This work proposes a hybrid classifier, based on one-class techniques, whose aim is to detect anomalous situations during the battery test. The faults are created by modifying the measured cell temperature a slight ratio from their real value. A detailed analysis of each technique performance is presented. The average performance of the chosen classifier presents successful results.