Hierarchical Support Vector Machine Research Articles

Cross-subject emotion recognition using hierarchical feature optimization and support vector machine with multi-kernel collaboration

Objective. Due to individual differences, it is greatly challenging to realize the multiple types of emotion identification across subjects. Approach. In this research, a hierarchical feature optimization method is proposed in order to represent emotional states effectively based on peripheral physiological signals. Firstly, sparse learning combined with binary search is employed to achieve feature selection of single signals. Then an improved fast correlation-based filter is proposed to implement fusion optimization of multi-channel signal features. Aiming at overcoming the limitations of the support vector machine (SVM), which uses a single kernel function to make decisions, the multi-kernel function collaboration strategy is proposed to improve the classification performance of SVM. Main results. The effectiveness of the proposed method is verified on the DEAP dataset. Experimental results show that the proposed method presents a competitive performance for four cross-subject types of emotion identification with an accuracy of 84% (group 1) and 85.07% (group 2). Significance. The proposed model with hierarchical feature optimization and SVM with multi-kernel function collaboration demonstrates superior emotion recognition accuracy compared to state-of-the-art techniques. In addition, the analysis based on DEAP dataset composition characteristics presents a novel perspective to explore the emotion recognition issue more objectively and comprehensively.

Research and Effectiveness Analysis of Innovative Teaching Models of Physical Education in Colleges and Universities under Convergent Network Architecture

Abstract In order to realize the innovation of sports teaching mode, this paper first combines the motion capture technology and three-dimensional modeling technology, based on the objectives and theories of sports teaching in colleges and universities, and constructs a college sports auxiliary teaching system under the fusion network architecture. Then, the three-dimensional modeling of sports action, the extraction of key sports action features, and the use of generalized discriminant analysis and hierarchical SVM construction method for the degradation and classification, respectively, to achieve the recognition of the sport in sports teaching. Finally, the effectiveness of the system and teaching model was tested and evaluated through the use of Loadrunner and SPSS software. The results show that the response time is 2s when the number of concurrent users is 100, the utilization rate of CPU is 64.2%, and the mastery of sports knowledge and movement skills of the experimental class is higher than that of the control class by 0.18 and 0.26, respectively. The innovative teaching mode of physical education in this study is effective.

A Dual-Optimization Fault Diagnosis Method for Rolling Bearings Based on Hierarchical Slope Entropy and SVM Synergized with Shark Optimization Algorithm.

Slope entropy (SlopEn) has been widely applied in fault diagnosis and has exhibited excellent performance, while SlopEn suffers from the problem of threshold selection. Aiming to further enhance the identifying capability of SlopEn in fault diagnosis, on the basis of SlopEn, the concept of hierarchy is introduced, and a new complexity feature, namely hierarchical slope entropy (HSlopEn), is proposed. Meanwhile, to address the problems of the threshold selection of HSlopEn and a support vector machine (SVM), the white shark optimizer (WSO) is applied to optimize both HSlopEn and an SVM, and WSO-HSlopEn and WSO-SVM are proposed, respectively. Then, a dual-optimization fault diagnosis method for rolling bearings based on WSO-HSlopEn and WSO-SVM is put forward. We conducted measured experiments on single- and multi-feature scenarios, and the experimental results demonstrated that whether single-feature or multi-feature, the WSO-HSlopEn and WSO-SVM fault diagnosis method has the highest recognition rate compared to other hierarchical entropies; moreover, under multi-features, the recognition rates are all higher than 97.5%, and the more features we select, the better the recognition effect. When five nodes are selected, the highest recognition rate reaches 100%.

Hierarchical Support Vector Machine based Classifier for Autocoding

This paper proposes a hierarchical support vector machine based classifier for autocoding. The purpose of this method is to improve classification accuracy utilizing numerical features obtained from probability scores by support vector machine (SVM). The proposed method captures the tendency to be incorrectly predicated for each label from the probability scores obtained from SVM. Using the captured information, we train new support vector machines for each targeted label considering the feature of the targeted label to improve the classification accuracy. The numerical examples with governmental survey data show a better performance of the proposed method.

Multiresolution hierarchical support vector machine for classification of large datasets

Multiresolution hierarchical support vector machine for classification of large datasets

FSM-HSVM-Based Locomotion Mode Recognition for Exoskeleton Robot

This paper proposes a hierarchical support vector machine recognition algorithm based on a finite state machine (FSM-HSVM) to accurately and reliably recognize the locomotion mode recognition of an exoskeleton robot. As input signals, this method utilizes the angle information of the hip joint and knee joint collected by inertial sensing units (IMUs) on the thighs and shanks of the exoskeleton and the plantar pressure information collected by force sensitive resistors (FSRs) are used as input signals. This method establishes a framework for mode transition by combining the finite state machine (FSM) with the common locomotion modes. The hierarchical support vector machine (HSVM) recognition model is then tightly integrated with the mode transition framework to recognize five typical locomotion modes and eight locomotion mode transitions in real-time. The algorithm not only reduces the abrupt change in the recognition of locomotion mode, but also significantly improves the recognition efficiency. To evaluate recognition performance, separate experiments are conducted on six subjects. According to the results, the average accuracy of all motion modes is 97.106% ± 0.955%, and the average recognition delay rate is only 25.017% ± 6.074%. This method has the benefits of a small calculation amount and high recognition efficiency, and it can be applied extensively in the field of robotics.

Anomaly detection of diabetes data based on hierarchical clustering and CNN

In this paper, we propose a diabetes data anomaly detection approach based on hierarchical clustering and support vector machine (SVM), named hierarchical support vector machine (HCSVM). In the HCSVM approach, the diabetes data sets with the same data characteristics are classified by clustering algorithm, and the data are divided into significant abnormal parts and potential abnormal parts. Additionally, the convolutional neural network (CNN) is utilized to detect and analyze the data of each part. The feature vector output from CNN full connection layer is applied as the input data of SVM classifier, and the optimal classification hyperplane is constructed in high-dimensional space for classification, so as to detect the abnormal data in diabetes data more pertinently. Finally, a real diabetes data set collected by a hospital is used for experiment, and ROC curve is adopted to evaluate the performance of the proposed approach compared with random forest algorithm, KNN algorithm and SVM algorithm. The results show that the HCSVM algorithm combined with hierarchical clustering and SVM can achieve a better performance.

Detection of Hepatitis B Virus-associated Hepatocellular Carcinoma Disease Using Hybrid Hierarchical k-Means Clustering and SVM Algorithm

Background: Hepatocellular carcinoma (HCC) is a serious disease and is the third main cause of death in the world. Hepatitis B virus infection can lead to HCC. The virus introduces genetic material into the host, damages DNA, and interferes with the activity of the apoptotic and tumor suppressors to trigger the formation of an oncogene. However, most of these cases are discovered after cancer enters stage three or four. Objective: Early detection of HCC through machine learning algorithm approach using data set: DNA sequence of HBx HepB virus. Methods: The research method used is the development of a Support Vector Machine classifier algorithm for carcinoma detection. The large data volume and unbalance data distribution in class can decrease the accuracy rate and sensitivity. Therefore, this paper proposed a hybrid of Hierarchical k-Means clustering and SVM algorithms to detect HCC disease using HBx DNA sequences. In this method, the SVM algorithm was applied in each cluster using the Hierarchical k- Means method. Results: The experimental result showed an accuracy rate of 97.18%, a sensitivity of 98.9%, and AUC of 0.918. This means the performance was increased to 9.52%, 95.3%, and 0.4 above the conventional SVM method. Conclusion: Detection of HCC can be applied using the SVM algorithm based on clustering. The proposed method, by hybrid hierarchical k-Means and SVM, increased the performance of classification results for the detection.

The recognition of plastic bottle using linear multi hierarchical SVM classifier

Both classification rate and accuracy are crucial for the recyclable PET bottles, and the existing combination methods of SVM all simply use SVM as the unit classifier, ignoring the improvement of SVM’s classification performance in the training process of deep learning. A linear multi hierarchical deep structure based on Support Vector Machine (SVM) is proposed to cover this problem. A novel definition of the input matrix in each layer enhances the optimization of Lagrange multipliers in Sequential Minimal Optimization (SMO) algorithm, thus the datapoint in maximum interval of SVM hyperplane could be recognized, improving the classification performance of SVM classifier in this layer. The loss function defined in this paper could control the depth of Linear Multi Hierarchical SVM (LMHSVM), the generalization parameters are added in the loss function and the input matrix to enhance the generalization performance of LMHSVM. The process of creating Bottle dataset by Histogram of Oriented Gradient (HOG) and Principal Component Analysis (PCA) is introduced meanwhile, reducing the data size of bottles. Experiments are conducted on LMHSVM and multiple typical classification algorithms with Bottle dataset and UCI datasets, the results indicated that LMHSVM has excellent classification performances than FNN classifier, LIBSVM (Gaussian) and GFS-AdaBoost-C in KEEL.

Alternative selection of Raman or LIBS spectral information in hierarchical discrimination of raw sapphires according to geographical origin for accuracy improvement

Both Raman spectroscopy and laser-induced breakdown spectroscopy (LIBS) were cooperatively utilized to improve the geographical origin identification of raw sapphires from five different countries (Mozambique, Laos, Australia, Rwanda, and Congo). A hierarchical support vector machine (H-SVM) was used for multi-group identification. Initially, accuracy improved to 87.5% using merged Raman-LIBS data compared to those of using only Raman (82.8%) or LIBS (71.9%) information. This improvement was attributed to incorporating two complimentary spectroscopic datasets that provided molecular vibrational and elemental information. However, merging both spectroscopic datasets is may not be the best choice since it would make distinct and sample-descriptive information in one spectroscopic dataset less recognized for analysis by the inclusion of less characteristic information in another spectroscopic dataset; using only Raman or LIBS information at each discrimination stage would be more effective. When Raman information was utilized during the first three discrimination stages followed by LIBS data during the fourth (last) discrimination stage in H-SVM, the accuracy improved to 90.6%. The proper selection of molecular vibrational or elemental sample information at different discrimination stages is attributed to this improvement.

Hierarchical support vector machine for facial micro-expression recognition

The sample category distribution of spontaneous facial micro-expression datasets is unbalanced, due to the experimental environment, collection equipment, and individualization of subjects, which brings great challenges to micro-expression recognition. Therefore, this paper introduces a micro-expression recognition model based on the Hierarchical Support Vector Machine (H-SVM) to reduce the interference of sample category distribution imbalance. First, we calculated the position of the apex frame in the micro-expression image sequence. To keep micro-expression frames balanced, we sparsely sample the images sequence according to the apex frame. Then, the Low-level Descriptors of the region of interest of the micro-expression image sequence and the High-level Descriptors of apex frame are extracted. Finally, the H-SVM model is used to classify the fusion features of different levels. The experimental results on SMIC, CAMSE2, SAMM, and their composite datasets show that our method can achieve superior performance in micro-expression recognition.

Abstract 1869: Methylation signatures associated with T790M status in progressive NSCLC

Abstract Background: Emergence of the EGFR T790M mutation accounts for acquired first generation EGFR TKI resistance in over half of patients with EGFR mutant NSCLC. In patients without emergent T790M, resistance mechanisms are less well understood. We explored the impact of DNA methylation status and TKI treatment failure in these patients. Methods: Using a prospective cohort of patients with acquired TKI resistance, tumour tissue samples pre/post TKI exposure were identified. DNA was extracted from FFPE tissue using the Qiagen AllPrep DNA/RNA FFPE Extraction Protocol, and subsequently analyzed using the Illumina Infinium EPIC array. Raw microarray data files were processed using the software package minfi for data normalization (Illumina method) and extraction of methylation levels (M-values). Samples were split into two groups according to the T790M status of each sample (T790M + or T790M-). The set of most informative probes, those whose M-value profiles align most closely with the T790M status of the study samples, was generated by selecting the 1,000 probes with lowest ANOVA's p-value. The stability of the resulting sample clustering was assessed by hierarchical clustering (Euclidean distance), classification with internal cross-validation (SVM leave-one-out), and non-parametric dimensional reduction (t-SNE). Results: 40 samples from 36 EGFR mutant NSCLC patients were successfully profiled. Pre TKI samples were available in 10 patients with an EGFR mutation of which 4 had matched post TKI tissue (3 T790M+, 1 T790M-). The remaining 26 samples in post TKI patients included 17 T790M + and 9 T790M- cases. A DNA methylation-based signature was developed by selecting the array probes that best discriminated T790M+ from T790M- cases. Group membership was stable, as shown by cross-validation by three different methods (hierarchical clustering, SVM leave-one-out and t-SNE). The 1,000 probe cut-off was arbitrarily selected; however, identical sample clusters were obtained using 500 or 2,000 methylation array probes. When analyzing the genomic location of the set of probes that form the signature, we found broad distribution across all chromosomes, thus, ruling out the possibility of selection bias due to focal or chromosome-level aberrations. Several genes contained a higher number of the selected probes, including EGFR, whose expression levels are known to be regulated at the methylation level in certain cancer types. Cluster analysis using the 1,000-probe signature revealed a high degree of concordance between EGFR T790M and DNA methylation status. All post-TKI (n=20) T790M+ samples concentrated within epi-group 2, whereas 8/10 T790M- samples were found within epi-group 1. Of the 4 patients with matched samples, 2 had baseline samples within epi-group 2 and went on to develop EGFR T790M post TKI. Of the 2 with baseline samples within epi-group 1, one went on to develop T790M (post-TKI epigroup 2) and one did not (post-TKI epigroup 1). Conclusions: We observed a concordance between T790M status and epi-group suggesting that the development of resistance to EGFR-TKIs may be associated with distinct DNA methylation signatures. This signature may be present at baseline and predict for subsequent emergence of T790M. Citation Format: Grainne M. O'Kane, Alberto J. León, Muqdas Shabir, Jennifer H. Law, Penelope A. Bradbury, Geoff Liu, Adrian Sacher, Frances A. Shepherd, Dax Torti, Tracy L. Stockley, Ming Tsao, Trevor J. Pugh, Natasha B. Leighl. Methylation signatures associated with T790M status in progressive NSCLC [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1869.

Semantic Classification of Images in Hierarchical Manner using Fuzzy Rules and HSVM Classifier

Hierarchical structure classification is more effective in concept based classification. The feature vector is varying with the different concepts, though it belongs to the same class. Images which are similar in appearance are not correlated semantically. Hence an author made an attempt to classify the images in the dataset in concept wise in a hierarchical manner. For this, an author manually clustered the same concept images in a layer by layer to explore the semantic correlation of the concepts. A number of nodes in the hierarchy tree represent the number of classes in the dataset. Each node is classified into sub-nodes by the fuzzy classifier using fuzzy association rules. Fuzzy inference system trained by the fuzzy rules base, which is generated by the fuzzy feature vector. The performance of the fuzzy classifier is validated with multiclass support vector machine classifier and hierarchical support vector machine classifier. The dataset used here is COREL dataset; an author selected 8 concepts for the classification. Keywords: Fuzzy classifier, fuzzy rule, hierarchical SVM, multiclass SVM, semantic concepts Cite this Article: Thirumala Lakshmi K, Usha Kingsly Devi K. Semantic Classification of Images in Hierarchical Manner using Fuzzy Rules and HSVM Classifier. Journal of Image Processing & Pattern Recognition Progress. 2020; 7(2): 33–54p.

Neural activities classification of left and right finger gestures during motor execution and motor imagery

ABSTRACT In this study, a new paradigm containing motor observation, motor execution, and motor imagery was designed to investigate whether motor imagery (MI) and motor execution (ME) of finger gestures can be used to extend commands of practical mBCIs. The subjects were instructed to perform or imagine 30 left and right finger gestures. Hierarchical support vector machine (hSVM) method was applied to classify four tasks (i.e., ME and MI tasks between left and right gestures). The average classification accuracies of motor imagery and execution tasks using fivefold cross-validation were 90.89 ± 9.87% and 74.08 ± 13.42% in first layer and second layer, respectively. The average accuracy of classification of four classes is 83.06 ± 7.29% overall. These results show that performing or imaging finger movements have the potential to extend the commands of the existing BCI, especially for healthy elderly living.

Hardware-Compliant Compressive Image Sensor Architecture Based on Random Modulations and Permutations for Embedded Inference

This work presents a compact CMOS Image Sensor (CIS) architecture enabling embedded object recognition facilitated by a dedicated end-of-column Compressive Sensing (CS), reducing on-chip memory needs. Our sensing scheme is based on a combination of random modulations and permutations leading to an implementation with very limited hardware impacts. It is designed to meet both theoretical ( i.e. , stable embedding, measurements incoherence) and practical requirements ( i.e. , silicon footprint, power consumption). The only additional hardware compared to a standard CIS architecture using first order incremental Sigma-Delta ( $\Sigma \Delta $ ) Analog to Digital Converter (ADC) are a pseudo-random data mixing circuit, an in- $\Sigma \Delta ~\pm 1$ modulator and a small Digital Signal Processor (DSP). On the algorithmic side, three variants are presented to perform the inference on compressed measurements with a tunable complexity ( i.e. , one-vs.-all SVM, hierarchical SVM and small ANN with 1-D max-pooling). An object recognition accuracy of $\simeq 98.8$ % is reached on the COIL database (COIL, 100 classes) using our dedicated Neural Network classifier. We stress that the signal-independent dimensionality reduction performed by our dedicated CS scheme (1/480 in $480 \times 640$ VGA resolution case) allows to dramatically reduce memory requirements mainly related to the remotely learned coefficients used for the inference stage.

Hierarchical Text Classification Using CNNs with Local Approaches

In this paper, we discuss the application of convolutional neural networks (CNNs) for hierarchical text classification using local top-down approaches. We present experimental results implementing a local classification per node approach, a local classification per parent node approach, and a local classification per level approach. A 20Newsgroup hierarchical training dataset with more than 20 categories and three hierarchical levels was used to train the models. The experiments involved several variations of hyperparameters settings such as batch size, embedding size, and number of available examples from the training dataset, including two variation of CNN model text embedding such as static (stat) and random (rand). The results demonstrated that our proposed use of CNNs outperformed flat CNN baseline model and both the flat and hierarchical support vector machine (SVM) and logistic regression (LR) baseline models. In particular, hierarchical text classification with CNN-stat models using local per parent node and local per level approaches achieved compelling results and outperformed the former and latter state-of-the-art models. However, using CNN with local per node approach for hierarchical text classification underperformed and achieved worse results. Furthermore, we performed a detailed comparison between the proposed hierarchical local approaches with CNNs. The results indicated that the hierarchical local classification per level approach using the CNN model with static text embedding achieved the best results, surpassing the flat SVM and LR baseline models by 7 % and 13 %, surpassing the flat CNN baseline by 5 %, and surpassing the h-SVM and h-LR models by 5 % and 10 %, respectively.

Fusion of Wearable and Contactless Sensors for Intelligent Gesture Recognition

A novel approach of fusing datasets from multiple sensors using a hierarchical support vector machine (HSVM) algorithm is presented. The validation of this method is experimentally carried out using an intelligent learning system that combines two different data sources. The sensors are based on a contactless sensor, which is a radar that detects the movements of the hands and fingers, as well as a wearable sensor, which is a flexible pressure sensor array that measures pressure distribution around the wrist. A HSVM architecture is developed to effectively fuse different data types in terms of sampling rate, data format, and gesture information from the pressure sensors and radar. In this respect, the proposed method is compared with the classification results from each of the two sensors independently. Herein, datasets from 15 different participants are collected and analyzed. The results show that the radar on its own provides a mean classification accuracy of 76.7%, whereas the pressure sensors provide an accuracy of 69.0%. However, enhancing the pressure sensors' output results with radar using the proposed HSVM algorithm improves the classification accuracy to 92.5%.

A Hierarchical SVM Based Behavior Inference of Human Operators Using a Hybrid Sequence Kernel

To train skilled unmanned combat aerial vehicle (UCAV) operators, it is important to establish a real-time training environment where an enemy appropriately responds to the action performed by a trainee. This can be addressed by constructing the inference method for the behavior of a UCAV operator from given simulation log data. Through this method, the virtual enemy is capable of performing actions that are highly likely to be made by an actual operator. To achieve this, we propose a hybrid sequence (HS) kernel-based hierarchical support vector machine (HSVM) for the behavior inference of a UCAV operator. Specifically, the HS kernel is designed to resolve the heterogeneity in simulation log data, and HSVM performs the behavior inference in a sequential manner considering the hierarchical structure of the behaviors of a UCAV operator. The effectiveness of the proposed method is demonstrated with the log data collected from the air-to-air combat simulator.

Bionic control of exoskeleton robot based on motion intention for rehabilitation training

ABSTRACTAs the complexity of obtaining irregular daily motion trajectory during upper limb rehabilitation training, this paper proposes the bionic control method for the presented exoskeleton robot arm based on motion intention. Firstly, the collected motion signal is pre-processing by filtering. Then the motion intention and motion mode of the processed signal are classified by using the hierarchical multi-classification support vector machine. Meanwhile, the adaptive Hopf oscillator network based on dynamic learning is used for offline learning of joint motion trajectory, and the parameters of the reproduced signal in different motion modes are obtained. Finally, the corresponding parameters are transferred according to the user’s intention, and the oscillator network is reconstructed to realize the periodic motion control of rehabilitation training. With experimental verification, the proposed method can follow the human body’s motion intention and reproduce the daily motion trajectory of the upper limb. The results show it can be used to conduct rehabilitation training for the patient.

Monitoring the level of hypnosis using a hierarchical SVM system.

Monitoring level of hypnosis is a major ongoing challenge for anesthetists to reduce anesthetic drug consumption, avoiding intraoperative awareness and prolonged recovery. This paper proposes a novel automated method for accurate assessing of the level of hypnosis with sevoflurane in 17 patients using the electroencephalogram signal. In this method, a set of distinctive features and a hierarchical classification structure based on support vector machine (SVM) methods, is proposed to discriminate the four levels of anesthesia (awake, light, general and deep states). The first stage of the hierarchical SVM structure identifies the awake state by extracting Shannon Permutation Entropy, Detrended Fluctuation Analysis and frequency features. Then deep state is identified by extracting the sample entropy feature; and finally light and general states are identified by extracting the three mentioned features of the first step. The accuracy of the proposed method of analyzing the brain activity during anesthesia is 94.11%; which was better than previous studies and also a commercial monitoring system (Response Entropy Index).