Selection Of Salient Features Research Articles

Towards reduced-complexity scene text recognition (RCSTR) through a novel salient feature selection

Towards reduced-complexity scene text recognition (RCSTR) through a novel salient feature selection

Unsupervised feature selection via dual space-based low redundancy scores and extended OLSDA

Spectral clustering is a widely used method for unsupervised feature selection (UFS) to generate pseudo labels. Nonetheless, it is acknowledged that graph algorithms suffer from issues such as redundancy and the dissatisfaction of connectivity, which greatly affect the quality of the learning local manifold. Moreover, existing UFS methods usually ignore the linear dependency among features and the role of non-negative semantics of the predicted labels. Hence, a novel algorithm using dual space-based low redundancy scores and extended orthogonal least square discriminant analysis (OLSDA), abbreviated as DLSEO, is proposed in this paper. Specifically, extended OLSDA is employed to derive non-negative clustering labels and a manifold structure, which avoids explicitly constructing a Laplacian graph. The dual space-based low redundancy scores related to mutual information eliminate redundant data and features, preventing their interference in the feature selection process. Moreover, ℓ2,1−2-norm is introduced to simultaneously guarantee the feature weight matrix's sparsity and ensure the selection of salient features. The results of experiments conducted on twelve benchmark datasets compared with several relevant methods show the superiority of DLSEO.

Fusion of Features: A Technique to Improve Autism Spectrum Disorder Detection Using Brain MRI Images

Autism Spectrum Disorder (ASD) is a major incident neurological disorder. Medical practitioners use different diagnostic techniques such as Electroencephalogram (EEG) Analysis, Magnetic Resonance Imaging (MRI) analysis, and traditional Behavioral Analysis for ASD detection. However, diagnosis success largely depends on specialists' knowledge and remains seldom accessible to remote patients. To address this issue, recently, various machine learning (ML) approaches have been developed for ASD detection using brain MRI images. The performance of these approaches is often limited because of poor feature discrimination, inferior quality of features, high feature length, and poor correlation of features. Thus, there is a need for robust feature extraction and selection techniques to improve the performance of ASD detection. The proposed work demonstrates a fusion of three features, namely Gray Level Co-occurrence Matrix (GLCM) based holistic texture features, Local Binary Pattern (LBP) based local texture features, and Geometrical Features of the Corpus Callosum (GFCC) from brain MRI images. Further, a correlation-based feature selection technique is employed for the salient feature selection from the GLCM, LBP, and GFCC set to improve the feature quality. The effectiveness of the selected feature is evaluated using three ML classifiers such as K-Nearest neighbor (KNN), Support Vector Machine (SVM), and Classification Tree (CT). The proposed ASD detection scheme provides an accuracy of 95.86% with 10-fold cross-validation with a CT classifier. It is observed that the accuracy of the proposed system is improved by 11.32% over the recent GLCM-based ASD system. The correlation-based feature selection techniques minimize the recognition time by 34.95% over the ASD system without feature selection.

Small samples-oriented intrinsically explainable machine learning using Variational Bayesian Logistic Regression: An intensive care unit readmission prediction case for liver transplantation patients

Recently, machine learning has revolutionized medical diagnosis and prediction, rivaling human experts in accuracy. However, the limited interpretability poses challenges in meeting stringent medical decision standards. To facilitate interpretable predictive inference, this article proposes a small samples-oriented intrinsically explainable machine learning model, an Optimal Variational Bayesian Logistic Regression (OVBLR) model with a salient feature estimation strategy (OVBLR-SFE). OVBLR-SFE sequentially learns from limited labeled samples, mining disease-related risk factors to provide intrinsic interpretability to its predictive inference automatically. The model achieves this by reasonably approximating the posterior probability through the proposed OVBLR model to estimate the feature weight and influential effect on model outputs to facilitate salient feature selection. The effectiveness and superiority of OVBLR-SFE are demonstrated through extensive experiments on four benchmark medical datasets from the UCI repository, including the Bone Marrow Transplant: Children Data Set, Breast Cancer Wisconsin (Original) Data Set, SPECT Heart Data Set, and Heart Disease Data Set, as well as a real-world case of intensive care unit readmission prediction. Experimental results reveal that the OVBLR-SFE can achieve high accuracies of 92.02%, 97.21%, 86.71%, and 84.47%, respectively, with an average classification accuracy of 90.10% stably on the four UCI datasets. Especially, OVBLR-SFE can accurately infer significant features that align with that obtained by the post-hoc interpretation approaches: SHAP and AcME, demonstrating remarkable interpretability. In the real application, the proposed method achieved a readmission prediction accuracy of 88.11% in predicting the readmission probabilities of liver transplantation patients. Notably, the inferred salient features, such as total Intensive Care Unit length of stay (ICUTime), N-terminal pro-brain natriuretic peptide(NTproBNP), Operating time of surgery(OperationTime_h), and intraoperative red blood cell transfusion (RBCT), emerges as crucial in readmission prediction, aligning with the assessments of clinical experts. The source code of this study can be accessed at https://github.com/xqw42/OVBLR-SFE.

Attention induced multi-head convolutional neural network for human activity recognition

Deep neural networks, including convolutional neural networks (CNNs), have been widely adopted for human activity recognition in recent years. They have attained significant performance improvement over traditional techniques due to their strong feature representation capabilities. Some of the challenges faced by the HAR community is the non-availability of a substantial amount of labeled training samples, and the higher computational cost and system resources requirements of deep learning architectures as opposed to shallow learning algorithms. To address these challenges, we propose an attention-based multi-head model for human activity recognition (HAR). This framework contains three lightweight convolutional heads, with each head designed using one-dimensional CNN to extract features from sensory data. The lightweight multi-head model is induced with attention to strengthen the representation ability of CNN, allowing for automatic selection of salient features and suppress unimportant ones. We conducted ablation studies and experiments on two publicly available benchmark datasets: WISDM and UCI HAR, to evaluate our model. The experimental outcome demonstrates the effectiveness of the proposed framework in activity recognition and achieves better accuracy while ensuring computational efficiency.

Sparse Adaptive Graph Convolutional Network for Leg Agility Assessment in Parkinson's Disease.

Motor disorder is a typical symptom of Parkinson's disease (PD). Neurologists assess the severity of PD motor symptoms using the clinical rating scale, i.e., MDS-UPDRS. However, this assessment method is time-consuming and easily affected by the perception difference of assessors. In the recent outbreak of coronavirus disease 2019, telemedicine for PD has become extremely urgent for clinical practice. To solve these problems, we developed an automated and objective assessment method of the leg agility task in the MDS-UPDRS using videos and a graph neural network. In this study, a sparse adaptive graph convolutional network (SA-GCN) was proposed to achieve fine-grained quantitative assessment of skeleton sequences extracted from videos. Specifically, the sparse adaptive graph convolutional unit with a prior knowledge constraint was proposed to perform adaptive spatial modeling of physical and logical dependency for skeleton sequences, thus achieving the sparse modeling of the discriminative spatial relationships. Subsequently, a temporal context module was introduced to construct the remote context dependency in the temporal dimension, hence determining the global changes of the task. A multi-domain attention learning module was also developed to integrate the static spatial features and dynamic temporal features, and then to emphasize the salient feature selection in the channel domain, thereby capturing the multi-domain fine-grained information. Finally, the evaluation results using a dataset with 148 patients and 870 samples confirmed the effectiveness and reliability of our scheme, and the method outperformed other related state-of-the-art methods. Our contactless method provides a new potential tool for automated PD assessment and telemedicine.

A framework for on-implant spike sorting based on salient feature selection

On-implant spike sorting methods employ static feature extraction/selection techniques to minimize the hardware cost. Here we propose a novel framework for real-time spike sorting based on dynamic selection of features. We select salient features that maximize the geometric-mean of between-class distances as well as the associated homogeneity index effectively to best discriminate spikes for classification. Wave-shape classification is performed based on a multi-label window discrimination approach. An external module calculates the salient features and discrimination windows through optimizing a replica of the on-implant operation, and then configures the on-implant spike sorter for real-time online operation. Hardware implementation of the on-implant online spike sorter for 512 channels of concurrent extra-cellular neural signals is reported, with an average classification accuracy of ~88%. Compared with other similar methods, our method shows reduction in classification error by a factor of ~2, and also reduction in the required memory space by a factor of ~5.

CNN-based salient features in HSI image semantic target prediction

ABSTRACTDeep networks have escalated the computational performance in the sensor-based high dimensional imaging such as hyperspectral images (HSI), due to their informative feature extraction competency. Therefore in this work, we have extracted the informative features from different CNN models for the benchmark HSI datasets. The deep features have concatenated with spectral features to increase the informative knowledge in the image datacube. The feature concatenation has massively increased the size of datacube. Therefore, we have applied an unsupervised maximum object identification-based salient feature selection to identify the most informative features of datacube and discard the less informative features to reduce the computational time without compromising the accuracy. It is an unsupervised feature selection approach that transforms the data into scale space and achieved robust and strong features. In the previous CNN-based methods, raw features have directly fed to the MLP (multilayer perception) layers for target prediction whereas we have provided our salient features into a multi-core SVM-based set-up and have achieved high accuracy with low computational time as compared to the previous state-of-art techniques.

Whale optimization algorithm-based email spam feature selection method using rotation forest algorithm for classification

Email has continued to be an integral part of our lives and as a means for successful communication on the internet. The problem of spam mails occupying a huge amount of space and bandwidth, and the weaknesses of spam filtering techniques which includes misclassification of genuine emails as spam (false positives) are a growing challenge to the internet world. This research work proposed the use of a metaheuristic optimization algorithm, the whale optimization algorithm (WOA), for the selection of salient features in the email corpus and rotation forest algorithm for classifying emails as spam and non-spam. The entire datasets were used, and the evaluation of the rotation forest algorithm was done before and after feature selection with WOA. The results obtained showed that the rotation forest algorithm after feature selection with WOA was able to classify the emails into spam and non-spam with a performance accuracy of 99.9% and a low FP rate of 0.0019. This shows that the proposed method had produced a remarkable improvement as compared with some previous methods.

An effective feature selection method based on maximum class separability for fault diagnosis of ball bearing

The paper deals with the development of a novel feature selection approach for bearing fault diagnosis to overcome drawbacks of the distance evaluation technique (DET); one of the well-established feature selection approaches. Its drawbacks are the influence of its effectiveness by the noise and the selection of salient features regardless of the classification system. To overcome these shortcomings, an optimal discrete wavelet transform (DWT) is firstly used to decompose the bearing vibration signal at different decomposition depths to enhance the signal to noise ratio. Then, a combination of DET with binary particle swarm optimisation (BPSO) algorithm and a criterion based on scatter matrices employed as an objective function are suggested to improve the classification performances and to reduce the computational time. Finally, support vector machine is utilised to automate the identification of different bearing conditions. From the obtained results, the effectiveness of the suggested method is proven.

An effective feature selection method based on maximum class separability for fault diagnosis of ball bearing

The paper deals with the development of a novel feature selection approach for bearing fault diagnosis to overcome drawbacks of the distance evaluation technique (DET); one of the well-established feature selection approaches. Its drawbacks are the influence of its effectiveness by the noise and the selection of salient features regardless of the classification system. To overcome these shortcomings, an optimal discrete wavelet transform (DWT) is firstly used to decompose the bearing vibration signal at different decomposition depths to enhance the signal to noise ratio. Then, a combination of DET with binary particle swarm optimisation (BPSO) algorithm and a criterion based on scatter matrices employed as an objective function are suggested to improve the classification performances and to reduce the computational time. Finally, support vector machine is utilised to automate the identification of different bearing conditions. From the obtained results, the effectiveness of the suggested method is proven.

Cervical cancer classification from Pap-smears using an enhanced fuzzy C-means algorithm

Abstract Globally, cervical cancer ranks as the fourth most prevalent cancer affecting women. However, it can be successfully treated if detected at an early stage. The Pap smear is a good tool for initial screening of cervical cancer, but there is the possibility of error due to human mistake. Moreover, the process is tedious and time-consuming. The objective of this study was to mitigate the risk of mistake by automating the process of cervical cancer classification from Pap smear images. In this research, contrast local adaptive histogram equalization was used for image enhancement. Cell segmentation was achieved through a Trainable Weka Segmentation classifier, and a sequential elimination approach was used for debris rejection. Feature selection was achieved using simulated annealing integrated with a wrapper filter, while classification was achieved using a fuzzy c-means algorithm. The evaluation of the classifier was carried out on three different datasets (single cell images, multiple cell images and Pap smear slide images from a pathology unit). An overall classification accuracy, sensitivity and specificity of ‘98.88%, 99.28% and 97.47%‘, ‘97.64%, 98.08% and 97.16%’ and ‘96.80%, 98.40% and 95.20%’ were obtained for each dataset respectively. The higher accuracy and sensitivity of the classifier was attributed to the robustness of the feature selection method that was utilized to select cell features that would improve the classification performance, and the number of clusters used during defuzzification and classification. The evaluation and testing conducted confirmed the rationale of the approach taken, which is based on the premise that the selection of salient features embeds sufficient discriminatory information that leads to an increase in the accuracy of cervical cancer classification. Results show that the method outperforms many of the existing algorithms in terms of the false negative rate (0.72%), false positive rate (2.53%), and classification error (1.12%), when applied to the DTU/Herlev benchmark Pap smear dataset. The approach articulated in this paper is applicable to many Pap smear analysis systems, but is particularly pertinent to low-cost systems that should be of significant benefit to developing economies.

A feature selection technique for generation of classification committees and its application to categorization of laryngeal images

This paper is concerned with a two phase procedure to select salient features (variables) for classification committees. Both filter and wrapper approaches to feature selection are combined in this work. In the first phase, definitely redundant features are eliminated based on the paired t-test. The test compares the saliency of the candidate and the noise features. In the second phase, the genetic search is employed. The search integrates the steps of training, aggregation of committee members, selection of hyper-parameters, and selection of salient features into the same learning process. A small number of genetic iterations needed to find a solution is the characteristic feature of the genetic search procedure developed. The experimental tests performed on five real-world problems have shown that significant improvements in classification accuracy can be obtained in a small number of iterations if compared to the case of using all the features available.