Multiclass Support Vector Machine Algorithm Research Articles

Wavelet Scattering Transform based Doppler signal classification

Normal blood supply to the human brain may be marred by the presence of a clot inside the blood vessels. This clot structure called emboli inhibits normal blood flow to the brain. It is considered as one of the main sources of stroke. Presence of emboli in human’s can be determined by the analysis of transcranial Doppler signal. Different signal processing and machine learning algorithms have been used for classifying the detected signal as an emboli, Doppler speckle, and an artifact. In this paper, we sought to make use of the wavelet transform based algorithm called Wavelet Scattering Transform, which is translation invariant and stable to deformations for classifying different Doppler signals. With its architectural resemblance to Convolutional Neural Network, Wavelet Scattering Transform works well on small datasets and subsequently was trained on a dataset consisting of 300 Doppler signals.To check the effectiveness of extracted Scattering transform based features for Doppler signal classification, learning algorithms that included multi-class Support vector machine, k-nearest neighbor and Naive Bayes algorithms were trained. Comparative analysis was done with respect to the handcrafted Continuous wavelet transform features extracted from samples and Wavelet scattering with Support vector machine achieved an accuracy of 98.89%. Also, with set of extracted scattering coefficients, Gaussian process regression was performed and a regression model was trained on three different sets of scattering coefficients with zero order scattering coefficients providing least prediction loss of 34.95%.

Combining surface‐enhanced Raman spectroscopy and mid‐infrared spectroscopy in a data fusion model to forensic differentiate the electrostatic copying paper

AbstractThe aim of this paper was to explore the non‐destructive application of fusion spectra model for characterization and differentiation of electrostatic copying papers that could be favourable to give forensic aid in legal cases. Two hundred fifty electrostatic copy paper samples were collected from various markets. All samples were subjected to surface‐enhanced Raman spectroscopy (SERS) analysis from 3500 to 400 cm−1 Raman shift range and attenuated total reflection‐Fourier transform infrared spectrum (ATR‐FTIR) analysis from 4000 to 400 cm−1 wavenumber range, respectively. The spectral data refracted the constituents present in the electrostatic copy papers were cellulose, inorganic filler calcium carbonate and barium sulfate. Fisher discriminant analysis (FDA), multi‐class support vector machine (MSVM) and decision tree (DT) algorithms were used to build the model. The precision rate, recall rate, F‐score and total accuracy were considered as indicators to evaluate the model's performance. The results showed that fusion models were superior to single model. The feature layer fusion model based on MSVM algorithm gave a differentiating power of 100% by grouping all the sample groups. This study demonstrated that spectral fusion model is a feasible and reliable approach for fast and non‐destructive differentiation of electrostatic copying papers. It has great potential in real application scenarios.

Joint Angle-Based Activity Recognition System for PARO Therapy Observation

PARO social robot has been introduced as an alternative therapy for dementia people. However, actual therapy implementation should be evaluated to improve the effectiveness of therapy. We, therefore, propose a system that able to observe and analyze the activity interaction between caregiver and patient in PARO therapy. In this paper, we focus on developing Efficient human activity recognition. The system uses features of the multi-camera-based human skeleton as the input information. We classify the activity recognition process into lower-body and upper-body recognition, which use 11 and 12 joint angles input, respectively. To extract the input data, the system uses a spatiotemporal method of multiclass Support Vector Machine (SVM) algorithm. We use the ADASYN approach for resampling imbalanced datasets. The proposed model has been tested in a sensor room, which utilizes 6 Kinect Azure cameras, to demonstrate its effectiveness. The result shows the proposed method succeeded in recognizing seven basic therapy activities. Then, the proposed method improved the accuracy of the conventional method based on therapeutic scenarios. Furthermore, the therapy's observation method of activity interaction will be conducted.

Incremental multiclass open-set audio recognition

Incremental learning aims to learn new classes if they emerge while maintaining the performance for previously known classes. It acquires useful information from incoming data to update the existing models. Open-set recognition, however, requires the ability to recognize examples from known classes and reject examples from new/unknown classes. There are two main challenges in this matter. First, new class discovery: the algorithm needs to not only recognize known classes but it must also detect unknown classes. Second, model extension: after the new classes are identified, the model needs to be updated. Focusing on this matter, we introduce incremental open-set multiclass support vector machine algorithms that can classify examples from seen/unseen classes, using incremental learning to increase the current model with new classes without entirely retraining the system. Comprehensive evaluations are carried out on both open set recognition and incremental learning. For open-set recognition, we adopt the openness test that examines the effectiveness of a varying number of known/unknown labels. For incremental learning, we adapt the model to detect a single novel class in each incremental phase and update the model with unknown classes. Experimental results show promising performance for the proposed methods, compared with some representative previous methods.

Aircraft Engine Audio Signal Analysis in Assisting Maintenance Inspections

As the core component of modern commercial aircraft, turbofan engines have long been the center of focus in aircraft maintenance. Being subject to high temperatures and immense pressures causes problems for the engine components, such as the fan blades, as they are frequently burdened with the potential of overhaul and malfunction. Over many years, the industry has seen various methods of engine inspection and maintenance, ranging from manual inspection to computing large quantities of pre-existing data. Within, audio signal analysis has stood out as a productive, non-invasive method, with many alternate studies analyzing sound signals from components such as the combustion chamber. However, many of these methods, despite demonstrating good accuracy, are incredibly complex and require sophisticated apparatus. Therefore, this study begins by investigating the sound generation process of turbofan engines, especially how the features and form of the fan blade characterize its audio signals. This investigation proposes a solution that utilizes a fast Multi-Class Support Vector Machine (SVM) algorithm based on fan-blade-related audio signals from a perspective similar to the classification of music and images through supervised machine learning. Experimental results show that this fast Multi-Class SVM is more effective than traditional machine learning methods in its accuracy, F1-score, and other indicators.

A fast automatic detection and classification of voltage magnitude anomalies in distribution network systems using PMU data

Timely anomaly detection and classification in voltage signals for distribution systems allows the design of preventive and corrective actions to avoid damage or loss of equipment. In this paper, an approach that combines a robust recurrence quantification analysis (RRQA) for features’ extraction that allows anomaly detection and classification through a multiclass support vector machine (SVM) algorithm is proposed. This approach is robust to noise, is free a filtering stage, has a low computational burden and is easy to interpret, becoming viable for online monitoring of distribution systems. For its validation, case studies with the presence of voltage magnitude anomalies (VMA) events that occur during normal operation conditions are analyzed. Thus, synthetic records generated through a Monte Carlo model are compared with other algorithms based on similar strategies. Finally, the proposed approach is assessed using PMU records installed in a distribution system to show its performance in a real world environment.

Exploring photoacoustic spectroscopy-based machine learning together with metabolomics to assess breast tumor progression in a xenograft model ex vivo

In the current study, a breast tumor xenograft was established in athymic nude mice by subcutaneous injection of the MCF-7 cell line and assessed the tumor progression by photoacoustic spectroscopy combined with machine learning tools. The advancement of breast tumors in nude mice was validated by tumor volume kinetics and histopathology and corresponding image analysis by TissueQuant software compared to controls. The ex vivo tumors in progressive conditions belonging to time points, day 5th, 10th, 15th & 20th, were excited with 281 nm pulsed laser light and recorded the corresponding photoacoustic spectra in time domain. The spectra were then pre-processed, augmented for a 10-fold increase in the data strength, and subjected to wavelet packet transformation for feature extraction and selection using MATLAB software. In the present study, the top 10 features from all the time point groups under study were selected based on their prediction ranking values using the mRMR algorithm. The chosen features of all the time-point groups were then subjected to multi-class Support Vector Machine (SVM) algorithms for learning and classifying into respective time point groups under study. The analysis demonstrated accuracy values of 95.2%, 99.5%, and 80.3% with SVM- Radial Basis Function (SVM-RBF), SVM-Polynomial & SVM-Linear, respectively. The serum metabolomic levels during tumor progression complemented photoacoustic patterns of tumor progression, depicting breast cancer pathophysiology.The current study reports the assessment of tumor progression in athymic nude mice by Photoacoustic spectroscopy-based machine learning tools. The progressive tumors were classified using multi-class Support Vector Machine (SVM) algorithms with 99.5% accuracy. The serum metabolomic levels during tumor progression complemented photoacoustic spectral features, depicting breast cancer pathophysiology.

Multi class SVM algorithm with active learning for network traffic classification

With the current massive amount of traffic that is going through the internet, internet service providers (ISPs) and networking service providers (NSPs) are looking for various ways to accurately predict the application type of flow that is going through the internet. Such prediction is critical for security and network monitoring applications as they require application type to be known in prior. Traditional ways using port-based or payload-based analysis are not sufficient anymore as many applications start using dynamic unknown port numbers, masquerading, and encryption techniques to avoid being detected. Recently, machine learning has gained significant attention in many prediction applications including traffic classification from flow features or characteristics. However, such algorithms suffer from an imbalanced data problem where some applications have fewer flow data and hence difficult to predict. In this paper, we employ network flow-level characteristics to identify the application type of traffic. Furthermore, we propose the use of an improved support vector machine (SVM) algorithm, named cost-sensitive SVM (CMSVM), to solve the imbalance problem in network traffic identification. CMSVM adopts a multi-class SVM algorithm with active learning which dynamically assigns a weight for applications. We examine the classification accuracy and performance of the CMSVM algorithm using two different datasets, namely MOORE_SET and NOC_SET datasets. Our results show that the CMSVM algorithm can reduce computation cost, improve classification accuracy and solve the imbalance problem when compared to other machine learning techniques.

Early Detection and Classification Approach for Plant Diseases based on MultiScale Image Decomposition

This paper presents a new and powerful approach for detecting and classifying leaf diseases for plant diagnosis with high accuracy. The main contribution of this paper is that a hybrid approach is proposed by using the combination of Partial Differential Equations (PDE) based image decomposition, segmentation, feature extraction, features selection and classification aiming to improve the classification accuracy and provide an excellent diagnosis. The TV-L1 Total variation model is adopted to separate the original image into texture and object components. Segmentation will be done only on the object component. Then texture, color, vein and shape features are extracted and merged in a feature vector using the codebook method. Moreover, features are selected by the RelieF feature selection algorithm to keep only relevant ones. In the classification, only selected features will be used and passed to the Multiclass Support Vector Machine algorithm SVM. The proposed approach is implemented and tested on the PV Plant Village dataset and provided a good and greater classification accuracy compared with the existing approaches from the literature. The obtained results proved that the use of PDE influences on the segmentation, which in turn, allowed us to identify correctly the leaves and provide new and optimal features, those features improves the classification accuracy rate to 95.9%.

Anomaly Detection and Bottleneck Identification of The Distributed Application in Cloud Data Center using Software–Defined Networking

Cloud computing applications have grown rapidly in the last decade, where many organizations are migrating their applications to cloud data center as they expected high performance, reliability, and the best quality of service. Data centers deploy a variety of technologies, such as software-defined networks (SDN), to effectively manage their resources. The SDN approach is a highly flexible network architecture that automates network configuration using a centralized controller to overcome traditional network limitations. This paper proposes an SDN-based monitoring algorithm to detect the performance anomaly and identify the bottleneck of the distributed application in the cloud data center using the support vector machine algorithm. It collects the data from the network devices and calculates the performance metrics for the distributed application components that are used to train the SVM algorithm and build a baseline model of the normal behavior of the distributed application. The SVM model detects performance anomaly behavior and identifies the root cause of bottlenecks using one-class support vector machine (OCSVM) and multi-class support vector machine (MCSVM) algorithms. The proposed method does not require any knowledge about the running applications or depends on static threshold values for performance measurements. Simulation results show that the proposed method can detect and locate the failure occurrences efficiently with high precision and low overhead compared to statistical methods, Naive Bayes Classifier and the decision tree machine learning method.

Detecting Drowsy Learners at the Wheel of e-Learning Platforms With Multimodal Learning Analytics

Learners are expected to stay wakeful and focused while interacting with e-learning platforms. Although wakefulness of learners strongly relates to educational outcomes, detecting drowsy learning behaviors only from log data is not an easy task. In this study, we describe the results of our research to model learners’ wakefulness based on multimodal data generated from heart rate, seat pressure, and face recognition. We collected multimodal data from learners in a blended course of informatics and conducted two types of analysis on them. First, we clustered features based on learners’ wakefulness labels as generated by human raters and ran a statistical analysis. This analysis helped us generate insights from multimodal data that can be used to inform learner and teacher feedback in multimodal learning analytics. Second, we trained machine learning models with multiclass-Support Vector Machine (SVM), Random Forest (RF) and CatBoost Classifier (CatBoost) algorithms to recognize learners’ wakefulness states automatically. We achieved an average macro-F1 score of 0.82 in automated user-dependent models with CatBoost. We also showed that compared to unimodal data from each sensor, the multimodal sensor data can improve the accuracy of models predicting the wakefulness states of learners while they are interacting with e-learning platforms.

Authorisation, attack detection and avoidance framework for IoT devices

Internet of Things (IoT) involve large volumes of data generated from the interactions between devices and people, and security is a main alarm in IoT. Most of the anomaly detection techniques in IoT use supervised machine learning technique which involve huge overhead and high false positives. It is observed that severity of attack response was not considered. In this study, the authors propose to develop an authorisation, attack detection and avoidance framework for IoT devices. Initially, traffic collection agent continuously gathers packet level and flow level information for a given time interval. Then detection agent (DA) first checks the collected information with the attack rules table. If any matching attack pattern is found, it informs the attack type to response agent (RA). On the other hand, if no matching pattern is found, then the classification agent applies multi-class support vector machine algorithm. Once the RA obtains the attack type from DA, then it estimates the severity of attack by computing the attack frequency over different time windows and appropriate action will be performed. Experimental results show that the proposed framework reduces 13% of unauthorised access and 19% false positive rate thereby increasing the detection accuracy by 0.6% and throughput.

Implementation of Multiclass Support Vector Machine for Classification of New Students Receiving Achievement Scholarships at Universitas Muhammadiyah Yogyakarta

The selection process for scholarship grantees at Universitas Muhammadiyah Yogyakarta (UMY) still utilizes the conventional method, namely Microsoft Excel. It is conducted by inputting all student data, then sorted from the highest to the lowest. Scholarships for prospective new students must be right on the target, meaning that they meet the criteria for students eligible for scholarships. It is intended not to harm other prospective students who should get scholarships. In previous studies, the classification process used many algorithms such as Naive Bayes, C.45, Decision Tree, k-Nearest Neighbor, and Support Vector Machine. The use of the Support Vector Machine algorithm employed a two-class classification. Support Vector Machine and Decision Tree algorithms are two classification methods that can obtain precise and accurate results. This study aims to use the Multiclass Support Vector Machine (LibSVM) algorithm to classify the achievement scholarship rankings for new students. The minimum amount of data used affected the classification results. From the 2015 to 2019 data, the highest amount of data was 2015, obtaining the highest accuracy result of 84.34% using the sigmoid kernel type and the k-fold value of 3. The classification was based on the entry system stages. PMDK stage 2 obtained an accuracy of 81.38%, with the most data amounting to 268 from stage 3.

Ensemble Generalized Multiclass Support-Vector-Machine-Based Health Evaluation of Complex Degradation Systems

Accurate health evaluation is crucial to reliable operation of complex degradation systems. Although traditional machine learning methods such as artificial neural network (ANN) and support vector machine (SVM) have been used widely, state assessment schemes based on a single classification model still suffer from low multiclass classification efficiency, high variance, and deviation. To solve these problems, this article proposes a novel health evaluation method based on stacking ensemble learning and generalized multiclass support vector machine (GMSVM) algorithm. The proposed health evaluation framework includes three parts: 1) abnormal value elimination and missing value processing are applied for multiple sensor data; 2) statistical features are extracted from the observed data and the Pearson correlation coefficient is applied for feature selection; and 3) ensemble generalized multiclass support vector machines (EGMSVMs) are utilized to evaluate the health situation of a degradation system. Unlike the binary classifiers and deep-learning-based classifiers, EGMSVMs utilize the stacking-based method to combine several GMSVMs as submodels and random forest as a metamodel, and the metamodel ensembles the results of submodels to reach a satisfied performance. Compared to traditional SVM- and ANN-based algorithms, EGMSVMs, in processing multiclass problems, achieve high efficiency and, meanwhile, low variance and deviation. The proposed method is verified using a hydraulic test rig. The experimental results show the feasibility and applicability of the proposed health evaluation framework.

A real-time FPGA accelerated stream processing for hyperspectral image classification

Realizing real-time classification of hyperspectral imagery has attracted researchers across various scientific and engineering disciplines. Several time-critical applications of high-resolution hyperspectral imaging require analysis of hyperspectral imagery with minimal turnaround time. In this regard, field-programmable gate arrays (FPGAs) provide cost-effective performance over the other high-performance reconfigurable computing systems. Developing such efficient systems poses several technical challenges such as designing and validating the FPGA architecture. A rapid prototyping approach offers several benefits in designing and developing FPGA architectures. This paper proposes a rapid prototyping method for the design and implementation of a multi-class support vector machine (MSVM) algorithm for real-time hyperspectral imagery classification using a low-cost FPGA system. We have evaluated design performance in terms of overall accuracy, resource utilization as well as timing requirements using four different datasets containing airborne, drone, and ground-based hyperspectral imagery. Experimental results show that the proposed FPGA design performs classification under strict real-time constraints.

Klasifikasi Kematangan Stroberi Berbasis Segmentasi Warna dengan Metode HSV

Classification of Strawberry Maturity Based on Color Segmentation using HSV Method. Manual fruit maturity classification has many limitations because it is influenced by human subjectivity. Hence, the application of digital image processing and artificial intelligence becomes more effective and efficient. This study aims to create a classification system that automatically divides strawberry maturity into three categories, namely not ripe, half-ripe, and ripe. The process of identifying the level of fruit maturity is based on the color characteristics Red, Green, Blue (RGB) value of the image. The method used for color segmentation is Hue, Saturation, Value (HSV) and for the classification of strawberry maturity using the Multi-Class Support Vector Machine (SVM) algorithm with a Radial Basic Function (RBF) kernel. Strawberry image data was retrieved using the Logitech C920 camera. The dataset consisted of 158 images of strawberries. The results showed that the classification of strawberry maturity using the multi-class SVM algorithm with kernel parameters RBF cost (C) = 10 and gamma (γ) = 10-3 produced the highest accuracy of 97%.

A multilayer perceptron and multiclass support vector machine based high accuracy technique for daily rainfall estimation from MSG SEVIRI data

The current paper introduces a new multilayer perceptron (MLP) and support vector machine (SVM) based approach to improve daily rainfall estimation from the Meteosat Second Generation (MSG) data. In this study, the precipitation is first detected and classified into convective and stratiform rain by two MLP models, and then four multi-class SVM algorithms were used for daily rainfall estimation. Relevant spectral and textural input features of the developed algorithms were derived from the spectral MSG SEVIRI radiometer channels. The models were trained using radar rainfall data set colected over north Algeria. Validation of the proposed daily rainfall estimation technique was performed by rain gauge network data set recorded over north Algeria. Thus, several statistical scores were calculated, such as correlation coefficient (r), root mean square error (RMSE), mean error (Bias), and mean absolute error (MAE). The findings given by: (r = 0.97, bias = 0.31 mm, RMSE = 2.20 mm and MAE = 1.07 mm), showed a quite satisfactory relationship between the estimation and the respective observed daily precipitation. Moreover, the comparison of the results with those of two advanced techniques based on random forests (RF) and weighted ‘k’ nearest neighbor (WkNN) showed higher accuracy obtained by the proposed model.

Rapidly and exactly determining postharvest dry soybean seed quality based on machine vision technology

The development of machine vision-based technologies to replace human labor for rapid and exact detection of agricultural product quality has received extensive attention. In this study, we describe a low-rank representation of jointly multi-modal bag-of-feature (JMBoF) classification framework for inspecting the appearance quality of postharvest dry soybean seeds. Two categories of speeded-up robust features and spatial layout of L*a*b* color features are extracted to characterize the dry soybean seed kernel. The bag-of-feature model is used to generate a visual dictionary descriptor from the above two features, respectively. In order to exactly represent the image characteristics, we introduce the low-rank representation (LRR) method to eliminate the redundant information from the long joint two kinds of modal dictionary descriptors. The multiclass support vector machine algorithm is used to classify the encoding LRR of the jointly multi-modal bag of features. We validate our JMBoF classification algorithm on the soybean seed image dataset. The proposed method significantly outperforms the state-of-the-art single-modal bag of features methods in the literature, which could contribute in the future as a significant and valuable technology in postharvest dry soybean seed classification procedure.

Experimental fault diagnosis for known and unseen operating conditions of centrifugal pumps using MSVM and WPT based analyses

Centrifugal pumps (CPs) are extensively used in various industries and their consistent operation ensures the keep-up of the plant’s work-flow. CPs most commonly fail due to mechanical component malfunction/breakdown and/or abnormalities in the fluid flow patterns. But these mechanical/hydraulic CP faults may not always be devoid of one another. Also, their coexistence may accelerate the CP failure. Hence, in the present study, both individually existing and coexisting hydraulic and mechanical faults with varying severities are simulated. On the whole, thirty three distinct faults of the CP system are considered. Motor current and CP vibration signatures are utilized to monitor the faults. In addition, the wavelet packet transform (WPT) analysis is used to pre-process the time-domain signal data in order to take into consideration the transients and spectral frequencies produced because of the complex fluid-structure interactions of mechanical and hydraulic faults. The faults classification is aimed at serving two purposes, (i) segregating different fault families and (ii) identifying the severity of a particular fault. Multiclass support vector machine algorithm (MSVM) is employed in classification of the fault data. Two methodologies based on best energy (BE) criterion and principal component analysis (PCA), that is, respectively, the WPT-BE-MSVM and WPT-PCA-MSVM are proposed and their performances are compared. The feature combinations giving the highest performance for each of these methodologies are identified. The algorithms are tested for wide operating conditions of the CP. An intermediate speed fault classification case is also studied to accommodate for the absence of data (unseen data case) at a particular operating speed of the CP. The classification accuracies obtained from both the methodologies are promising.

Online Diagnostics of Mechanical and Electrical Faults in Induction Motor Using Multiclass Support Vector Machine Algorithms Based on Frequency Domain Vibration and Current Signals

This paper demonstrates the development of a flexible fault diagnosis methodology that can detect up to ten different faults in the induction motor (IM), simultaneously. The major IM electrical faults, such as the broken rotor bar (BRB), phase unbalance (PUF), and stator winding fault (SWF), and mechanical faults, such as bearing fault (BF), unbalanced rotor (UR), bowed rotor (BR), and misaligned rotor (MR), are considered with different fault severities for the diagnosis. The experiments are conducted with three varying loads and seven different speeds, and the frequency domain vibration and current data are acquired at a relatively low sampling rate of 1 kHz. Several statistical features are extracted and then the best feature-set is selected using the wrapper model. Thereafter, a data classification tool based on the support vector machine (SVM) is used for the fault characterization. Initially, a multi-fault diagnosis is performed by training and testing the SVM at the same operating conditions (i.e., load and speed). The performance of the classifier is found to be very good at all IM operating conditions. The main focus of this study lies in overcoming the fault diagnosis, where the data are unavailable at required operating conditions. This is accomplished by employing interpolation and extrapolation strategies for different loads and speeds. The proposed methodology not only solves practical problem of unavailability of data at different operating conditions but also shows good performance and takes low computation time, which are vital requirements of an online intelligent condition monitoring system.