On One-class Support Vector Machine Research Articles

Fault Diagnosis of Marine Turbocharger System Based on an Unsupervised Algorithm

The fault diagnosis of a marine turbocharger system is very crucial for realizing intelligent operation and maintenance in a big data analysis context. In order to improve the diagnostic rate of faults in engineering applications, in this paper, a new unsupervised machine learning algorithm, which is based on one-class support vector machine (OSVM), affinity propagation (AP) and Gaussian mixture model (GMM), called OAGFD is proposed for fault diagnosis. OSVM was firstly used to divide samples of marine turbocharger system into normal and fault samples, and only the fault samples are used in following steps to identify specific fault types. The AP was adopted automatically to provide an initial value for expectation maximization, which can obtain the maximum value of iteration parameters. The GMM is used to classify faults of marine turbocharger system and output the fault diagnosis results. Finally, the OAGFD is validated by actual data. The experiment results show that OAGFD can quickly and accurately be trained. The OAGFD method can achieve higher identification accuracy for multi-faults of marine turbocharger system and takes on faster operation speed and stronger generalization ability than tradition methods. It is an efficient and unsupervised fault diagnosis technique and has both theoretical and practical value. This research provides a new method for automatic fault diagnosis of the marine turbocharger system.

MAPPING OF BURNED AREA USING PRESENCE AND BACKGROUND LEARNING FRAMEWORK ON THE GOOGLE EARTH ENGINE PLATFORM

Abstract. Mapping of burned areas caused by forest fires was always a main concern to researchers in the field of remote sensing. Thus, various spectral indices and classification techniques have been proposed in the literature. In such a problem, only one specific class is of real interest and could be referred to as a one-class classification problem. One-class classification methods are highly desirable for quick mapping of classes of interest. A common used solution to deal with One-Class classification problem is based on oneclass support vector machine (OC-SVM). This method has proved useful in classification of remote sensing images. However, overfitting problem and difficulty in tuning parameters have become the major obstacles for this method. The new Presence and Background Learning (PBL) framework does not require complicated model selection and can generate very high accuracy results. On the other hand the Google Earth Engine (GEE) portal provides access to satellite and other ancillary data, cloud computing, and algorithms for processing large amounts of data with relative ease. Therefore, this study mainly aims to investigate the possibility of using the PBL framework within the GEE platform to extract burned areas from freely available Landsat archive in the year 2015. The quality of the results obtained using PBL framework was assessed using ground truth digitized by qualified technicians and compared to other classification techniques: Thresholding burned area spectral Index (BAI) and OC-SVM classifiers. Experimental results demonstrate that PBL framework for mapping the burned areas shows the higher classification accuracy than the other classifiers, and it highlights the suitability for the cases with few positive labelled samples available, which facilitates the tedious work of manual digitizing.

A Class-Incremental Learning Method Based on One Class Support Vector Machine

Class-incremental learning refers to the problem that the number of classes increases dynamically in the training stage. In this paper, a method based on one class support vector machine (OCSVM) is proposed for class incremental learning. The basic idea of the proposed method is that one class model is used to learn the distribution of new class, and then the knowledge learned and previous models are used for class incremental learning. The proposed method does not need to store all training data and makes full use of the learned information, so it has lower memory cost and time consumption. Also, the proposed method can build model when only one class is available. Compared with OC2MC, CIL, NN, the proposed method also achieves competitive results on different datasets.

An anomaly detection system based on variable N-gram features and one-class SVM

Context: Run-time detection of system anomalies at the host level remains a challenging task. Existing techniques suffer from high rates of false alarms, hindering large-scale deployment of anomaly detection techniques in commercial settings.Objective: To reduce the false alarm rate, we present a new anomaly detection system based on a novel feature extraction technique, which combines the frequency with the temporal information from system call traces, and on one-class support vector machine (OC-SVM) detector.Method: The proposed feature extraction approach starts by segmenting the system call traces into multiple n-grams of variable length and mapping them to fixed-size sparse feature vectors, which are then used to train OC-SVM detectors.Results: The results achieved on a real-world system call dataset show that our feature vectors with up to 6-grams outperform the term vector models (using the most common weighting schemes) proposed in related work. More importantly, our anomaly detection system using OC-SVM with a Gaussian kernel, trained on our feature vectors, achieves a higher-level of detection accuracy (with a lower false alarm rate) than that achieved by Markovian and n-gram based models as well as by the state-of-the-art anomaly detection techniques.Conclusion: The proposed feature extraction approach from traces of events provides new and general data representations that are suitable for training standard one-class machine learning algorithms, while preserving the temporal dependencies among these events.

Anomalous Human Activity Detection Based on Online One-Class SVM with N-Grams Kernel

We present a novel online unsupervised anomaly detection method for human activities. The proposed approach is based on one-class support vector machine (OCSVM) clustering, where the novelty detection SVM capabilities are used for the identification of anomalous activities. Particular attention is given to activity classification in absence of a priori information on the distribution of outliers. Activities are represented by variable-length event sequences, but the most commonly used kernels are defined on fixed-dimension spaces. To solve the problem, we develop a novel sequence-similarity kernel, the n-grams kernel. Our kernel is conceptually simple and efficient to compute and performs well in comparison with state-of-the-art methods for anomaly detection. Moreover, most SVM algorithms require large number of memory to store the kernel matrix, or repeated access to the training samples. This makes it infeasible for online anomaly detection. In this paper, we develop simple and computationally efficient online learning algorithms for anomaly detection.

Audio classification based on one-class SVM

The author studied an audio classification method based on One-Class Support Vector Machine(OCSVM),which could form a decision function for every single class sample and accordingly obtain the aim of classification based on maximum of decision function.By employing wavelet packed transformation to extract features of audio and integrating multiple features,five audio classes were made: pure speech,music,environmental sound,speech over background and silence.Experimental results show that OCSVM has better classification accuracy,and performs better than the other classification systems using the Bayes,Hidden Markov Model(HMM) and Neural Network(NN).