Directed Acyclic Graph Support Vector Machine Research Articles

An Intelligent Detection Method for Obstacles in Agricultural Soil with FDTD Modeling and MSVMs

Unknown objects in agricultural soil can be important because they may impact the health and productivity of the soil and the crops that grow in it. Challenges in collecting soil samples present opportunities to utilize Ground Penetrating Radar (GPR) image processing and artificial intelligence techniques to identify and locate unidentified objects in agricultural soil, which are important for agriculture. In this study, we used finite-difference time-domain (FDTD) simulated models to gather training data and predict actual soil conditions. Additionally, we propose a multi-class support vector machine (MSVM) that employs a semi-supervised algorithm to classify buried object materials and locate their position in soil. Then, we extract echo signals from the electromagnetic features of the FDTD simulation model, including soil type, parabolic shape, location, and energy magnitude changes. Lastly, we compare the performance of various MSVM models with different kernel functions (linear, polynomial, and radial basis function). The results indicate that the FDTD-Yee method enhances the accuracy of simulating real agricultural soils. The average recognition rate of the hyperbola position formed by the GPR echo signal is 91.13%, which can be utilized to detect the position and material of unknown and underground objects. For material identification, the directed acyclic graph support vector machine (DAG-SVM) model attains the highest classification accuracy among all soil layers when using an RBF kernel. Overall, our study demonstrates that an artificial intelligence model trained with the FDTD forward simulation model can effectively detect objects in farmland soil.

Railway Track Recognition Based on Radar Cross-Section Statistical Characterization Using mmWave Radar

The track settlement has a great influence on the safe operation of high-speed trains. The existing track settlement measurement approach requires sophisticated or expensive equipments, and the real-time performance is limited. To address the issue, an ultra-high resolution track settlement detection method is proposed by using millimeter wave radar based on frequency modulated continuous wave (FMCW). Firstly, by constructing the RCS statistical feature data set of multiple objects in the track settlement measurement environment, a directed acyclic graph-support vector machine (DAG-SVM) based method is designed to solve the problem of track recognition in multi-object scenes. Then, the adaptive chirp-z-transform (ACZT) algorithm is used to estimate the distance between the radar and the track surface, which realizes automatic real-time track settlement detection. An experimental platform has been constructed to verify the effectiveness of the proposed method. The experimental results show that the accuracy of track classification and identification is at least 95%, and the accuracy of track settlement measurement exceeds 0.5 mm, which completely meets the accuracy requirements of the railway system.

A vision-based fall detection framework for the elderly in a room environment using motion features and DAG-SVM

Falls are one of the major health risks for the elderly in public health care issues, thus fall detection is particularly necessary for the seniors. In this paper, we propose a novel computer vision-based fall activity detection framework which is based on a new feature extraction method of human motion. We detect falls by analyzing the motion process of human body rather than conventional human shape analysis. Approximate ellipse is first adopted to fit human silhouette, and the areas outside the ellipse are removed to confirm human body silhouette (HBS). Then HBS is divided into three regions, and the motion features are extracted from the centroid coordinate changes of each region. A directed acyclic graph support vector machine (DAG-SVM) classifier is finally applied to detect falls from normal daily activities. The advantage of the proposed method is that it can not only effectively distinguish falls from fall-like activities but also detect the falls parallel to camera optical axis. Experimental results show that our method achieves a state-of-the-art accuracy in fall detection. This method is lightweight and thus can be integrated into embedded devices for real-time health surveillance.

Facial Expression Recognition Using Kernel Entropy Component Analysis Network and DAGSVM

Facial expression recognition (FER) plays a significant part in artificial intelligence and computer vision. However, most of facial expression recognition methods have not obtained satisfactory results based on low-level features. The existed methods used in facial expression recognition encountered the major issues of linear inseparability, large computational burden, and data redundancy. To obtain satisfactory results, we propose an innovative deep learning (DL) model using the kernel entropy component analysis network (KECANet) and directed acyclic graph support vector machine (DAGSVM). We use the KECANet in the feature extraction stage. In the stage of output, binary hashing and blockwise histograms are adopted. We sent the final output features to the DAGSVM classifier for expression recognition. We test the performance of our proposed method on three databases of CK+, JAFFE, and CMU Multi-PIE. According to the experiment results, the proposed method can learn high-level features and provide more recognition information in the stage of training, obtaining a higher recognition rate.

A Hybrid Model Combining Learning Distance Metric and DAG Support Vector Machine for Multimodal Biometric Recognition

Metric learning has significantly improved machine learning applications such as face re-identification and image classification using K-Nearest Neighbor (KNN) and Support Vector Machine (SVM) classifiers. However, to the best of our knowledge, it has not been investigated yet, especially for the multimodal biometric recognition problem in immigration, forensic and surveillance applications with uncontrolled ear datasets. Therefore, it is interesting and very attractive to propose a novel framework for multimodal biometric recognition based on Learning Distance Metric (LDM) via kernel SVM. This paper considers metric learning for SVM by investigating a hybrid Learning Distance Metric and Directed Acyclic Graph SVM (LDM-DAGSVM) model for multimodal biometric recognition, where LDM and DAGSVM are two emerging techniques in dealing with classification problems. Different from existing multimodal biometric recognition methods, the proposed approach aims to learn Mahalanobis distance metric via kernel SVM to maximize the inter-class variations and minimize the intra-class variations, simultaneously. Experimental results on the uncontrolled datasets such as AR face and AWE ear datasets show that the proposed approach achieves competitive performance compared with models working on individual modalities and overperforms the state-of-the-art multimodal methods. The proposed model achieves five-fold classification accuracy around 99.85 % for the face and ear images.

An Advanced Genetic Algorithm with Improved Support Vector Machine for Multi-Class Classification of Real Power Quality Events

This article proposes an application of non-dominated sorting genetic algorithm III (NSGA III) and directed acyclic graph support vector machine (DAG-SVM) based combined approach for recognition and classification of power quality disturbances. Power disturbances are non-stationary and non-linear by nature and require a large number of features vectors for detection, which results in high computation time and error. The proposed method first employs NSGA III algorithm for an effective features extraction from power signals. The numbers of features required for detection are less in case of proposed NSGA III algorithm. The NSGA III algorithm generates optimal solutions based on multi objective optimization and then fitness function is generated with the help of Pareto front to obtain unique features set from power signals. The NSGA III not only has exact localization of power quality events and stronger robustness but also reduces the computational time in comparison with the traditional NSGA II. The obtained unique feature vectors are used for training of DAG-SVM to classify the power quality disturbances. In case of N-classes, the proposed DAG-SVM generates N(N-1)/2 classifiers, one for each pair of classes and makes the decision accurate and faster. The short event detection, lesser computational timing, superior classification accuracy, and high anti-noise performance are the main advantages of the proposed method. Furthermore, Virtex-5 FPGA based processor is used to test and validate the feasibility of the proposed method for real time analysis of power quality events.

Automatic identification and classification of power quality events using a hybrid intelligent approach

This article proposes an implementation of Enhanced-Non-dominated Sorting Genetic Algorithm (E-NSGA III) and Directed Acyclic Graph-Support Vector Machine (DAG-SVM) based combined approach for recognition and classification of power quality events. The function of E-NSGA III is to extract features and DAG-SVM for the purpose of classification of power quality events with minimum error. The non-stationary and non- linear nature of power quality disturbances makes it a suitable choice for E-NSGA III. This technique gives unique Pareto-optimal solutions based on multi-objective optimization. Considering equal priority for all the objectives, a fitness function is used to obtain the best solution set from the first Pareto front. Non-dominated sorting sets a dominated count and assigns domination to each individual. Dominated count means the number of times an individual data point has been dominated by another data point. The obtained unique feature vectors are used for training of DAG-SVM classifier to classify the power quality disturbances. Many power quality events including harmonics, swell, sag, transient, swell with harmonics and sag with harmonics are taken into consideration for investigation of power signal disturbances. Comparative investigation reveals the novelty and efficacy of the proposed hybrid technique in comparison of support vector machine, radial basis feed-forward neural network, probabilistic neural network, support vector machine and type-1 and type-II based fuzzy classifier under serve noise conditions. The simulation results show that the Directed Acyclic Graph Support Vector Machine gives the best classification accuracy of 99.20% and less computation time.

Dual Tree Complex Wavelet Transform with Multiobjective Optimization Algorithm for Real‐Time Power Quality Events Classification

AbstractThis article proposes an application of Dual Tree Complex Wavelet Transform (DTCWT) with non‐dominated sorting genetic algorithm III (NSGA III) based multi‐objective optimization and directed acyclic graph support vector machine (DAG‐SVM) for recognition and classification of power quality events. The proposed method first employs DTCWT for extraction of features then NSGA III algorithm for an effective optimization of extracted features. After that, DAG‐SVM based classifier are used to predict the classes of PQ disturbances. The NSGA III algorithm generates optimalsolutions based on multi objective optimization and then fitness function is generated with the help of Pareto front to obtain unique features set from power signals. The NSGA III not only optimizes selected features with DTCWT but also reduces the computational time in comparison with the traditional NSGA II. The obtained unique feature vectors are used for training of Directed Acyclic Graph‐SVM to classify the power quality disturbances. The short event detection, lesser computational timing, superior classification accuracy, and high anti‐noise performance are the main advantages of the proposed method.

Fast CU partition decision for H.266/VVC based on the improved DAG-SVM classifier model

One of the biggest changes in H.266/Versatile Video Coding (VVC) is introduced quad-tree with nested multi-type tree (QTMT) coding tree architecture, where the multi-type tree (MTT) structure in H.266/VVC includes binary tree (BT) and ternary tree (TT). Compared with H.265/High Efficiency Video Coding (HEVC) which only is divided by quad-tree (QT), the QTMT architecture makes the coding unit (CU) partition procedure more complexity. In this paper, we design a fast CU partition decision algorithm based on the improved Directed Acyclic Graph Support Vector Machine (DAG-SVM) model to reduce the complexity of CU partition. The video sequences are first encoded on the H.266/VVC and Test Model 4.0 (VTM 4.0), and the characteristics of the video sequences are extracted for training through the improved F-score method, where the correlation between a feature and CU partition is high. Then, the offline training is used for the improved DAG-SVM model. Finally, the trained DAG-SVM model is embedded in VTM 4.0 to early forecast the optimal CU partition modes. Simulation results indicate that the proposed method increases the time savings to 54.74% while maintaining the encoding performance. Furthermore, the proposed method exceeds the latest methods of H.266/VVC.

Monitoring tip-based nanomachining process by time series analysis using support vector machine

In this paper, time-series data analysis and pattern recognition using a multi-class support vector machine (SVM) were studied to monitor the state changes of the AFM tip-based nanomachining process with respect to the machining performance and tip wear. Time series data (i.e. machining force from the process), which has transient, nonlinear, and non-stationary characteristics, was collected by a data acquisition system. Three status detection features including the maximum force, peak-to-peak force value, and the variance of the collected lateral machining force, were extracted to classify the state of the nanomachining process. Directed Acyclic Graph Support Vector Machines (DAGSVM) with a Gaussian Radial Basis Kernel Function (RBF Kernel) was constructed to identify the different process states. Using this multi-class SVM, the machining process and the tip wear can be classified into three regions, which are effective machining with a sharp tip, transition region and bad/no machining with severe tip wear. The experimental data showed that the accuracy of the SVM was over 94.73% in both binary and ternary classifications, which confirmed that the SVM-based pattern recognition technology via time series data could successfully monitor the tip wear and process performance for tip-based nanomachining process.

Cross-Voting SVM Method for Multiple Vehicle Classification in Wireless Sensor Networks.

A novel multi-class classification method named the voting-cross support vector machine (SVM) method was proposed in this study, for classifying vehicle targets in wireless sensor networks. The advantages and disadvantages of available methods were summarized, after a comparative analysis of commonly used multi-objective classification algorithms. To improve the classification accuracy of multi-class classification and ensure the low complexity of the algorithm for engineering implementation on wireless sensor network (WSN) nodes, a framework was proposed for cross-matching and voting on the category to which the vehicle belongs after combining the advantages of the directed acyclic graph SVM (DAGSVM) method and binary-tree SVM method. The SVM classifier was selected as the basis two-class classifier in the framework, after comparing the classification performance of several commonly used methods. We utilized datasets acquired from a real-world experiment to validate the proposed method. The calculated results demonstrated that the cross-voting SVM method could effectively increase the classification accuracy for the classification of multiple vehicle targets, with a limited increase in the algorithm complexity. The application of the cross-voting SVM method effectively improved the target classification accuracy (by approximately 7%), compared with the DAGSVM method and the binary-tree SVM method, whereas time consumption decreased by approximately 70% compared to the DAGSVM method.

Facial emotion classification using concatenated geometric and textural features

Geometric and textural features have been separately used in the literature for facial emotion classification. In this paper, we theoretically and empirically study the capture of facial expressions through geometric and texture-based features, and demonstrate that a simple concatenation of these features can lead to significant improvement in facial emotion classification. We also propose the use of the Directed Acyclic Graph SVM (DAGSVM) for facial emotion classification using the concatenated feature, by analyzing DAGSVM structures. We perform experiments using the well-known extended Cohn-Kanade (CK+), the MUG facial expression (MUG) and the Japanese Female Facial Expression (JAFFE) databases to evaluate the integration of geometric and textural features, and the use of DAGSVM for facial emotion classification. The said integration is found to be effective and DAGSVM is found to be computationally efficient in facial emotion classification.

Market basket analysis from egocentric videos

This paper presents Visual Market Basket Analysis (VMBA), a novel application domain for egocentric vision systems. The final goal of VMBA is to infer the behavior of the customers of a store during their shopping. The analysis relies on image sequences acquired by cameras mounted on shopping carts. The inferred behaviors can be coupled with classic Market Basket Analysis information (i.e., receipts) to help retailers to improve the management of spaces and marketing strategies. To set up the challenge, we collected a new dataset of egocentric videos during real shopping sessions in a retail store. Video frames have been labeled according to a proposed hierarchy of 14 different customer behaviors from the beginning (cart picking) to the end (cart releasing) of their shopping. We benchmark different representation and classification techniques and propose a multi-modal method which exploits visual, motion and audio descriptors to perform classification with the Directed Acyclic Graph SVM learning architecture. Experiments highlight that employing multimodal representations and explicitly addressing the task in a hierarchical way is beneficial. The devised approach based on Deep Features achieves an accuracy of more than 87% over the 14 classes of the considered dataset.

Problem formulations and solvers in linear SVM: a review

Support vector machine (SVM) is an optimal margin based classification technique in machine learning. SVM is a binary linear classifier which has been extended to non-linear data using Kernels and multi-class data using various techniques like one-versus-one, one-versus-rest, Crammer Singer SVM, Weston Watkins SVM and directed acyclic graph SVM (DAGSVM) etc. SVM with a linear Kernel is called linear SVM and one with a non-linear Kernel is called non-linear SVM. Linear SVM is an efficient technique for high dimensional data applications like document classification, word-sense disambiguation, drug design etc. because under such data applications, test accuracy of linear SVM is closer to non-linear SVM while its training is much faster than non-linear SVM. SVM is continuously evolving since its inception and researchers have proposed many problem formulations, solvers and strategies for solving SVM. Moreover, due to advancements in the technology, data has taken the form of ‘Big Data’ which have posed a challenge for Machine Learning to train a classifier on this large-scale data. In this paper, we have presented a review on evolution of linear support vector machine classification, its solvers, strategies to improve solvers, experimental results, current challenges and research directions.

Fault Diagnosis of the Planetary Gearbox Based on ssDAG-SVM

Planetary gearbox is of great significance for many practical cases, and many data-driven approaches have been employed to solve the fault diagnosis problem for the system. Among these methods, Directed Acyclic Graph Support Vector Machines (DAG-SVM) has been widely adopted due to its ability to handle the multi-class problem. Different from traditional DAG-SVM, a structure-selected DAG-SVM (ssDAG-SVM) is proposed such that the diagnosis performance will not degrade because of inappropriate node structure. By introducing the concept of class separability, the principle of evaluating the degree of class separability is integrated into the process of constructing the DAG-SVM structure. Subsequently, a proper structure can be selected to realize the planetary gearbox fault diagnosis with high accuracy. Finally, the effectiveness of the method is illustrated by some practical experiments.

A Fast Classification Method of Faults in Power Electronic Circuits Based on Support Vector Machines

AbstractFault detection and location are important and front-end tasks in assuring the reliability of power electronic circuits. In essence, both tasks can be considered as the classification problem. This paper presents a fast fault classification method for power electronic circuits by using the support vector machine (SVM) as a classifier and the wavelet transform as a feature extraction technique. Using one-against-rest SVM and one-against-one SVM are two general approaches to fault classification in power electronic circuits. However, these methods have a high computational complexity, therefore in this design we employ a directed acyclic graph (DAG) SVM to implement the fault classification. The DAG SVM is close to the one-against-one SVM regarding its classification performance, but it is much faster. Moreover, in the presented approach, the DAG SVM is improved by introducing the method of Knearest neighbours to reduce some computations, so that the classification time can be further reduced. A rectifier and an inverter are demonstrated to prove effectiveness of the presented design.

Wavelet Entropy and Directed Acyclic Graph Support Vector Machine for Detection of Patients with Unilateral Hearing Loss in MRI Scanning.

Highlights We develop computer-aided diagnosis system for unilateral hearing loss detection in structural magnetic resonance imaging.Wavelet entropy is introduced to extract image global features from brain images. Directed acyclic graph is employed to endow support vector machine an ability to handle multi-class problems.The developed computer-aided diagnosis system achieves an overall accuracy of 95.1% for this three-class problem of differentiating left-sided and right-sided hearing loss from healthy controls. Aim: Sensorineural hearing loss (SNHL) is correlated to many neurodegenerative disease. Now more and more computer vision based methods are using to detect it in an automatic way.Materials: We have in total 49 subjects, scanned by 3.0T MRI (Siemens Medical Solutions, Erlangen, Germany). The subjects contain 14 patients with right-sided hearing loss (RHL), 15 patients with left-sided hearing loss (LHL), and 20 healthy controls (HC).Method: We treat this as a three-class classification problem: RHL, LHL, and HC. Wavelet entropy (WE) was selected from the magnetic resonance images of each subjects, and then submitted to a directed acyclic graph support vector machine (DAG-SVM).Results: The 10 repetition results of 10-fold cross validation shows 3-level decomposition will yield an overall accuracy of 95.10% for this three-class classification problem, higher than feedforward neural network, decision tree, and naive Bayesian classifier.Conclusions: This computer-aided diagnosis system is promising. We hope this study can attract more computer vision method for detecting hearing loss.

Detection and Classification of Power Quality Disturbances Using Double Resolution S-Transform and DAG-SVMs

The accurate detection and classification of power quality (PQ) disturbances in power systems is a key step to determine the causes of these events before any proper countermeasure could be taken. This paper presents a new algorithm for detection and classification of PQ disturbances based on the combination of double-resolution S-transform (DRST) and directed acyclic graph support vector machines (DAG-SVMs). The proposed method first employs DRST for an effective feature extraction from power signals. Then, the DAG-SVMs are used to predict the classes of PQ disturbances. The DRST not only has better time–frequency localization and stronger robustness but also reduces the computational complexity without losing the useful information of the original signal in comparison with the traditional S-transform. Through the combined use of DRST and DAG-SVMs, the algorithm can be easily implemented in embedded real-time applications. Finally, the implementation of the proposed algorithm in a digital signal processor + advanced reduced instruction set computing machine-based hardware test platform is introduced. The effectiveness of the proposed method is demonstrated by means of computer simulations and practical experiments with single and combined PQ disturbances.

Electromagnetic side-channel attack based on PSO directed acyclic graph SVM

Machine learning has a powerful potential for performing the template attack (TA) of cryptographic device. To improve the accuracy and time consuming of electromagnetic template attack (ETA), a multi-class directed acyclic graph support vector machine (DAGSVM) method is proposed to predict the Hamming weight of the key. The method needs to generate K(K1)/2 binary support vector machine (SVM) classifiers and realizes the K-class prediction using a rooted binary directed acyclic graph (DAG) testing model. Further, particle swarm optimization (PSO) is used for optimal selection of DAGSVM model parameters to improve the performance of DAGSVM. By exploiting the electromagnetic emanations captured while a chip was implementing the RC4 algorithm in software, the computation complexity and performance of several multi-class machine learning methods, such as DAGSVM, one-versus-one (OVO)SVM, one-versus-all (OVA)SVM, Probabilistic neural networks (PNN), K-means clustering and fuzzy neural network (FNN) are investigated. In the same scenario, the highest classification accuracy of Hamming weight for the key reached 100%, 95.33%, 85%, 74%, 49.67% and 38% for DAGSVM, OVOSVM, OVASVM, PNN, K-means and FNN, respectively. The experiment results demonstrate the proposed model performs higher predictive accuracy and faster convergence speed.

Fault detection and isolation for Polymer Electrolyte Membrane Fuel Cell systems by analyzing cell voltage generated space

This paper proposes a data-driven diagnostic approach for Polymer Electrolyte Membrane Fuel Cell (PEMFC) systems. Fault detection and isolation (FDI) is realized by analyzing individual cell voltages. A feature extraction method Fisher Discriminant Analysis (FDA) and a multi-class classification method Directed Acyclic Graph Support Vector Machine (DAGSVM) are utilized successively to extract the useful features from raw data and classify the extracted features into various classes related to health states. Experimental data of two different stacks are used to validate the proposed approach. The results show that five concerned faults can be detected and isolated with a high accuracy. Moreover, the light computational cost of the approach enhances the possibility of its online implementation.