Robust Feature Set Research Articles

Enhancing Online Security: A Novel Machine Learning Framework for Robust Detection of Known and Unknown Malicious URLs

The rapid expansion of the internet has led to a corresponding surge in malicious online activities, posing significant threats to users and organizations. Cybercriminals exploit malicious uniform resource locators (URLs) to disseminate harmful content, execute phishing schemes, and orchestrate various cyber attacks. As these threats evolve, detecting malicious URLs (MURLs) has become crucial for safeguarding internet users and ensuring a secure online environment. In response to this urgent need, we propose a novel machine learning-driven framework designed to identify known and unknown MURLs effectively. Our approach leverages a comprehensive dataset encompassing various labels—including benign, phishing, defacement, and malware—to engineer a robust set of features validated through extensive statistical analyses. The resulting malicious URL detection system (MUDS) combines supervised machine learning techniques, tree-based algorithms, and advanced data preprocessing, achieving a high detection accuracy of 96.83% for known MURLs. For unknown MURLs, the proposed framework utilizes CL_K-means, a modified k-means clustering algorithm, alongside two additional biased classifiers, achieving 92.54% accuracy on simulated zero-day datasets. With an average processing time of under 14 milliseconds per instance, MUDS is optimized for real-time integration into network endpoint systems. These outcomes highlight the efficacy and efficiency of the proposed MUDS in fortifying online security by identifying and mitigating MURLs, thereby reinforcing the digital landscape against cyber threats.

Developing a Dynamic Feature Selection System (DFSS) for Stock Market Prediction: Application to the Korean Industry Sectors

For several years, a growing interest among numerous researchers and investors in predicting stock price movements has spurred extensive exploration into employing advanced deep learning models. These models aim to develop systems capable of comprehending the stock market’s complex nature. Despite the immense challenge posed by the diverse factors influencing stock price forecasting, there remains a notable lack of research focused on identifying the essential feature set for accurate predictions. In this study, we propose a Dynamic Feature Selection System (DFSS) to predict stock prices across the 10 major industries, as classified by the FnGuide Industry Classification Standard (FICS) in South Korea. We apply 16 feature selection algorithms from filter, wrapper, embedded, and ensemble categories. Subsequently, we adjust the settings of industry-specific index data to evaluate the model’s performance and robustness over time. Our comprehensive results identify the optimal feature sets that significantly impact stock prices within each sector at specific points in time. By analyzing the inclusion ratios and significance of the optimal feature set by category, we gain insights into the proportion of feature classes and their importance. This analysis ensures the interpretability and reliability of our model. The proposed methodology complements existing methods that do not consider changes in the types of variables significantly affecting stock prices over time by dynamically adjusting the input variables used for learning. The primary goal of this study is to enhance active investment strategies by facilitating the creation of diversified portfolios for individual stocks across various sectors, offering robust models and feature sets that consistently demonstrate high performance across industries over time.

Improved temporal IoT device identification using robust statistical features.

The Internet of Things (IoT) is becoming more prevalent in our daily lives. A recent industry report projected the global IoT market to be worth more than USD 4 trillion by 2032. To cope with the ever-increasing IoT devices in use, identifying and securing IoT devices has become highly crucial for network administrators. In that regard, network traffic classification offers a promising solution by precisely identifying IoT devices to enhance network visibility, allowing better network security. Currently, most IoT device identification solutions revolve around machine learning, outperforming prior solutions like port and behavioural-based. Although performant, these solutions often experience performance degradation over time due to statistical changes in the data. As a result, they require frequent retraining, which is computationally expensive. Therefore, this article aims to improve the model performance through a robust alternative feature set. The improved feature set leverages payload lengths to model the unique characteristics of IoT devices and remains stable over time. Besides that, this article utilizes the proposed feature set with Random Forest and OneVSRest to optimize the learning process, particularly concerning the easier addition of new IoT devices. On the other hand, this article introduces weekly dataset segmentation to ensure fair evaluation over different time frames. Evaluation on two datasets, a public dataset, IoT Traffic Traces, and a self-collected dataset, IoT-FSCIT, show that the proposed feature set maintained above 80% accuracy throughout all weeks on the IoT Traffic Traces dataset, outperforming selected benchmark studies while improving accuracy over time by +10.13% on the IoT-FSCIT dataset.

GastroVRG: Enhancing early screening in gastrointestinal health via advanced transfer features

The accurate classification of endoscopic images is a challenging yet critical task in medical diagnostics, which directly affects the treatment and management of Gastrointestinal diseases. Misclassification can lead to incorrect treatment plans, adversely affecting patient outcomes. To address this challenge, our research aimed to develop a reliable computational model to improve the accuracy of classifying conditions of esophagitis and polyps. We focused on a subset of the Kvasir v1 secondary dataset, comprising 2000 endoscopic images evenly distributed across two classes: esophagitis and polyp. The goal was to leverage the strengths of both Machine Learning(ML) and Deep Learning(DL) to create a model that not only predicts with high accuracy but also integrates seamlessly into clinical workflows. To this end, we introduced a novel VRG-based ensemble image feature extraction technique, combining the powers of VGG, RF, and GB models to synthesize a robust feature set conducive to high-precision classification. The ensemble approach demonstrated a best-in-class performance with the GB model achieving an outstanding 99.73% accuracy in detecting esophagitis and polyps. The practical implications of these results are substantial, indicating that our method can significantly improve diagnostic accuracy in real-world settings, reduce the rate of misdiagnosis, and contribute to the efficient and effective treatment of patients, ultimately enhancing the quality of healthcare services. With the successful application of our proposed method to a controlled dataset, future work involves deploying the model in clinical environments and expanding its application to a broader spectrum of Gastrointestinal conditions across multi-class datasets.

Evaluating the association between radiomic features on breast MRI and dormant tumor cell presence in the SURMOUNT trial.

561 Background: While breast cancer treatment is effective at eradicating primary disease, approximately 20% of patients will ultimately recur at a distant site. Recurrence has been linked to the presence of minimal residual disease (MRD) after treatment, which presents as circulating micrometastases and dormant tumor cells (DTCs) located primarily in the bone marrow. Radiomic features on MRI are imaging biomarkers that have been linked with lesion heterogeneity, which in turn has been associated with disease aggressiveness. Here, we examine the association between radiomic features from the primary lesion on MRI and the presence of post-treatment DTCs. Noninvasively identifying patients at high risk of MRD would facilitate optimized treatment strategies to reduce recurrence risk. Methods: Dynamic contrast-enhanced MRI data from breast cancer survivors enrolled in the SURMOUNT trial (N=104) were retrospectively analyzed. Participants with MRIs collected within 90 days of diagnosis who consented to provide a bone marrow aspirate (BMA) sample were eligible. Working with a board-certified radiologist, we successfully identified and segmented the primary lesion for N=70 patients with BMA outcomes. Using the publicly available CaPTk software, we computed 33 radiomic features for each segmented lesion based on pixel intensity, morphology, and grey level textures. To reduce the dimensionality of the feature space, we used these feature values to group the participants into clusters—termed radiomic phenotypes — via agglomerative hierarchical clustering with significance testing. Fisher’s exact test was applied to test the statistical significance of the association between radiomic phenotype and BMA outcome (± DTC). For all statistical tests, p<0.05 was considered significant. Results: Of the N=70 participants analyzed, 58 were negative and 12 were positive for DTCs present in the bone marrow. Two statistically significant radiomic phenotypes ( p<0.05) were identified that were strongly associated with BMA outcome ( p=0.0251). 11 out of 12 positive outcomes were grouped into the second radiomic phenotype (Table). Conclusions: We found a statistically significant association between DTC presence and radiomic phenotypes derived from MRI data. Given the large percentage of positive outcomes assigned to the second radiomic phenotype, the radiomic phenotypes may indicate low and high risk of MRD, respectively. Importantly, this analysis shows promise for the ability to predict MRD from noninvasive imaging features. These results motivate continued data collection to increase sample size and identify a robust set of radiomic features, which may ultimately be used as predictors in potential models of MRD risk. [Table: see text]

Deriving and interpreting robust features for survival prediction of brain tumor patients

AbstractAccurate prediction of survival days (SD) is vital for planning treatments in glioma patients, as type‐IV tumors typically have a poor prognosis and meager survival rates. SD prediction is challenging and heavily dependent on the extracted feature sets. Additionally, comprehending the behavior of complex machine learning models is a vital yet challenging aspect, particularly to integrate such models into the medical domain responsibly. Therefore, this study develops a robust feature set and an ensemble‐based regressor model to predict patients' SD accurately. We aim to understand how these features behave and contribute to predicting SD. To accomplish this, we employed post‐hoc interpretable techniques, precisely Shapley Additive exPlanations (SHAP), Partial Dependence Plots (PDP), and Accumulated Local Effects (ALE) plots. Furthermore, we introduced an investigation to establish a direct connection between radiomic features and their biological significance to enhance the interpretability of radiomic features. The best SD scores on the BraTS2020 training set are 0.504 for accuracy, 59927.38 mean squared error (MSE), 20101.86 median squared error (medianSE), and 0.725 Spearman ranking coefficient (SRC). The validation set's accuracy is 0.586, MSE is 76529.43, medianSE is 41402.78, and SRC is 0.52. The proposed predictor model exhibited superior performance compared with leading contemporary approaches across multiple performance metrics.

A Generative Approach to Person Reidentification.

Person Re-identification is the task of recognizing comparable subjects across a network of nonoverlapping cameras. This is typically achieved by extracting from the source image a vector of characteristic features of the specific person captured by the camera. Learning a good set of robust, invariant and discriminative features is a complex task, often leveraging contrastive learning. In this article, we explore a different approach, learning the representation of an individual as the conditioning information required to generate images of the specific person starting from random noise. In this way we decouple the identity of the individual from any other information relative to a specific instance (pose, background, etc.), allowing interesting transformations from one identity to another. As generative models, we use the recent diffusion models that have already proven their sensibility to conditioning in many different contexts. The results presented in this article serve as a proof-of-concept. While our current performance on common benchmarks is lower than state-of-the-art techniques, the approach is intriguing and rich of innovative insights, suggesting a wide range of potential improvements along various lines of investigation.

Abstract A020: An early determination of patients eligibility for a bladder cancer immunotherapy using a data science approach

Abstract Advanced bladder cancer patients have limited treatment options and, in most cases, limited overall survival. The first clinical trial with adoptive T cell therapy (ACT) that uses bladder cancer patients’ autologous tumor-infiltrating T lymphocytes (TILs) has been recently opened at Moffitt. To help identify which patients could benefit from ACT-TIL we are developing a machine learning predictor that takes advantage of tumor evolution data, such as tumor stages and grades routinely collected in clinic. The first phase of ACT-TIL involves resection of the primary tumor and the growth of TILs followed by a rapid expansion protocol of the reactive T cells. It may take up to 4-5 weeks before bladder cancer patients can get their own expanded TILs. Our predictor aims to determine whether the tumor isolated from a patient will grow TILs. In this work, we present a machine learning predictor of expansion of tumor infiltrating lymphocytes (ML-PETIL) that uses retrospectively collected patient tumor data with either demographic, clinical, or biological sample-related features. The proposed method consists of a suite of algorithms to identify a robust set of predictive features (feature selection algorithms) that are used to discriminate the dataset into two classes-that with successful TIL expansion or with no TIL growth (classification algorithms). Our predictor includes metrics which are used to address data imbalance and handle small to medium-sized datasets that contain features of varying datatypes and magnitudes. The ML-PETIL protocol offers an early determination of patients who would be eligible for ACT-TIL therapy. Citation Format: Kayode D. Olumoyin, Ahmet M. Aydin, Brittany L Bunch, Shari Pilon-Thomas, Michael Poch, Katarzyna A. Rejniak. An early determination of patients eligibility for a bladder cancer immunotherapy using a data science approach [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr A020.

Embedded Feature Selection Approach Using Penalized Logistic Regression for Universal Steganalysis

For universal steganalysis, the Spatial Rich Model (SRM) provides a robust feature set in the form of 106 feature submodels (34671-dimensional) to attack highly secure steganography methods. Nonetheless, the curse of dimensionality emerges as a drawback, prompting researchers to explore feature selection methods to achieve optimal performance for steganalyzers. In previous work, a wrapper feature selection method using global and local best particle swarm optimization (GLBPSO) has been used to select 28 submodels of SRM out of 106 submodels. The 28 submodels contain a total of 9152 features and an ensemble classifier has been used for the classification results. In this study, an embedded feature selection and classification approach using logistic regression (LR) with an elastic net penalty has been proposed and applied to the 9152 features. The proposed approach efficiently selects an informative subset of 9152 features while training the LR classifier. This not only reduces the number of features by a significant 75% but also boosts the classification performance when contrasted with the ensemble classifier. The proposed approach achieves an accuracy of 87.41%, surpassing the current state-of-the-art steganalysis methods based on deep learning by a substantial margin of 3-7%. It also outperforms a recent machine learning-based steganalysis method by 13%.

Improved fetal heartbeat detection using pitch shifting and psychoacoustics

ObjectiveFetal phonocardiography is a cost-effective and non-invasive method for monitoring of fetal health. However, the captured signal is often too noisy and unstable for reliable analysis. Detection of fetal heartbeats therefore requires a robust set of features to differentiate relevant sounds from background noise. We propose a novel set of features based on pitch shifting and psychoacoustics which lead to significant improvements of automatic fetal heartbeat detection. MethodsWe collected, annotated, cleaned, processed and segmented microphone recordings taken from pregnant women. In addition, we employed a well-known simulated dataset to validate the methodology further. A total of 212 features were extracted, including classical audio features, psychoacoustic features, and features based on Empirical Mode Decomposition. Methods from statistical analysis and machine learning were applied to demonstrate the relevance of specific features. Afterwards, we trained a set of machine learning algorithms with different subsets of features. ResultsPsychoacoustic features extracted from pitch-shifted signals are generally shown to be relevant and useful features for fetal heartbeat detection. When psychoacoustic features were added to the baseline features, classification accuracy of a robust classifier significantly increased from 69.24% to 76.16% for the real-world signals, and from 92.69% to 97.42% for one of the simulated datasets. ConclusionThe results demonstrate considerable gains in detection accuracy once pitch shifting and psychoacoustic modeling are applied to the input signal. SignificanceImprovements achieved from applying psychoacoustics on fetal phonocardiographic signals might be an important finding in the effort of making remote monitoring of fetal wellbeing inexpensive and accessible.

Multimodal Kernel-based discriminant correlation analysis data-fusion approach: an automated autism spectrum disorder diagnostic system.

Autism spectrum disorder (ASD) diagnostic systems, based on association of multimodal tools such as combination of Electroencephalogram (EEG) and eye-tracking, have emerged as an analytical to provide objective biomarkers. However, the existing feature-redundancy-based systems have lacked in providing knowledge of fusion approaches and robust feature-set. The present paper aims to reduce disorder homogeneity by proposing a multimodal diagnostic system which can incorporate multimodal data. The paper has collected simultaneous-data from three modalities (laptop-performance tool, EEG machine, and Eye-tracker) fused the recorded computational, neural and visual data. The multimodal features are analyzed via proposed multimodal Kernel-based discriminant correlation analysis (MKDCA) fusion approach and classified using state-of-the-art machine-learning classifiers. The proposed framework has considered the distinct cardinality of the feature vectors and fused the group structure among multiple samples after ranking them in increasing order. As per the results, the proposed multimodal system provided fused feature set of 11 influential features out of total 39 features. The SVM classifier has diagnosed ASD with 92% testing accuracy and 0.988 AUC(ROC). The proposed automated fusion-based system has the potential to classify disorder by reducing the disorder heterogeneity and stratifying ASD individuals into homogeneous sub-groups. In future, the correlation of reduced feature set with ASD clinical symptoms accounted by screening scales can provide clinical relevance of proposed model.

Texture Analysis of Adenomatous and Metastatic Adrenal Lesions on Native and Contrast-enhanced Computed Tomography

Differentiating lipid-poor adrenal adenomas from adrenal metastases is not possible without performing a dedicated contrast-enhanced CT (CECT) protocol. Our purpose is to evaluate the possible role of texture analysis in classifying adrenal lesions.
 This is a retrospective study. We evaluated 33 patients with 47 metastases and 43 patients with 47 adenomas. Seven of the adenomas were lipid-poor, and the rest – lipid-rich adenomas. We used commercially available software for lesion segmentation and texture analysis on native and arterial phase CT images with a slice thickness of 2 mm. The segmentation was semi-automated, and features were computed on the resulting regions of interest (ROI) for each lesion on both native and arterial phases. Two conventional (HU standard deviation, Histogram Entropy) and six second-order texture features (GLCM – Homogeneity, Energy, Entropy log 10, Contrast, Dissimilarity, NGDLM–Busyness) were calculated. For statistical analysis the IBM SPSS 19 package was used to compare the following groups in both phases: 1. Adenomas--Metastasis; 2. Lipid-rich adenomas (LRA)--metastases; 3. Lipid-rich--lipid-poor adenomas (LPA); 4. Lipid-poor adenomas--metastases.
 There was a statistically significant difference in the distribution of some conventional (Histogram Entropy) and second-order features (GLCM – Homogeneity, Energy, Entropy log 10, Dissimilarity) in the first group on the native as well as on the enhanced images. The results were similar when comparing lipid-rich adenomas to metastases. Only native phase derived features were discriminative between lipid-rich and lipid-poor adenomas, with no difference between parameters on CECT. No difference was found between any of the texture features in lipid-poor adenomas and metastases for both phases.
 First- and second-order texture features were graded based on their potential for serving as classifier tools. Unenhanced features ranked higher. Further research and validation are needed to discover the most robust set of features for differentiating between lipid-poor adenomas and metastases.

MethSemble-6mA: an ensemble-based 6mA prediction server and its application on promoter region of LBD gene family in Poaceae.

The Lateral Organ Boundaries Domain (LBD) containing genes are a set of plant-specific transcription factors and are crucial for controlling both organ development and defense mechanisms as well as anthocyanin synthesis and nitrogen metabolism. It is imperative to understand how methylation regulates gene expression, through predicting methylation sites of their promoters particularly in major crop species. In this study, we developed a user-friendly prediction server for accurate prediction of 6mA sites by incorporating a robust feature set, viz., Binary Encoding of Mono-nucleotide DNA. Our model,MethSemble-6mA, outperformed other state-of-the-art tools in terms of accuracy (93.12%). Furthermore, we investigated the pattern of probable 6mA sites at the upstream promoter regions of the LBD-containing genes in Triticum aestivum and its allied species using the developed tool. On average, each selected species had four 6mA sites, and it was found that with speciation and due course of evolution in wheat, the frequency of methylation have reduced, and a few sites remain conserved. This obviously cues gene birth and gene expression alteration through methylation over time in a species and reflects functional conservation throughout evolution. Since DNA methylation is a vital event in almost all plant developmental processes (e.g., genomic imprinting and gametogenesis) along with other life processes, our findings on epigenetic regulation of LBD-containing genes have dynamic implications in basic and applied research. Additionally, MethSemble-6mA (http://cabgrid.res.in:5799/) will serve as a useful resource for a plant breeders who are interested to pursue epigenetic-based crop improvement research.

Radiomics: A radiological evidence-based artificial intelligence technique to facilitate personalized precision medicine in hepatocellular carcinoma.

The high postoperative recurrence rates in hepatocellular carcinoma (HCC) remain a major hurdle in its management. Appropriate staging and treatment selection may alleviate the extent of fatal recurrence. However, effective methods to preoperatively evaluate pathophysiologic and molecular characteristics of HCC are lacking. Imaging plays a central role in HCC diagnosis and stratification due to the non-invasive diagnostic criteria. Vast and crucial information is hidden within image data. Other than providing a morphological sketch for lesion diagnosis, imaging could provide new insights to describe the pathophysiological and genetic landscape of HCC. Radiomics aims to facilitate diagnosis and prognosis of HCC using artificial intelligence techniques to harness the immense information contained in medical images. Radiomics produces a set of archetypal and robust imaging features that are correlated to key pathological or molecular biomarkers to preoperatively risk-stratify HCC patients. Inferred with outcome data, comprehensive combination of radiomic, clinical and/or multi-omics data could also improve direct prediction of response to treatment and prognosis. The evolution of radiomics is changing our understanding of personalized precision medicine in HCC management. Herein, we review the key techniques and clinical applications in HCC radiomics and discuss current limitations and future opportunities to improve clinical decision making.

New Ways for Patients to Make Sense of Their Electronic Health Record Data Using the Discovery Web Application: Think-Aloud Evaluation Study.

In the United States, patients can access their electronic health record (EHR) data through patient portals. However, current patient portals are mainly focused on a single provider, with very limited data sharing capabilities and put low emphasis on independent sensemaking of the EHR data. This makes it very challenging for patients to switch between different portals and aggregate the data to obtain a complete picture of their medical history and to make sense of it. Owing to this fragmentation, patients are exposed to numerous inconveniences such as medical errors, repeated tests, and limited self-advocacy. To overcome the limitations of EHR patient portals, we designed and developed Discovery-a web-based application that aggregates EHR data from multiple providers and present them to the patient for efficient exploration and sensemaking. To learn how well Discovery meets the patients' sensemaking needs and what features should such applications include, we conducted an evaluation study. We conducted a remote study with 14 participants. In a 60-minute session and relying on the think-aloud protocol, participants were asked to complete a variety of sensemaking tasks and provide feedback upon completion. The audio materials were transcribed for analysis and the video recordings of the users' interactions with Discovery were annotated to provide additional context. These combined textual data were thematically analyzed to surface themes that reflect how participants used Discovery's features, what sensemaking of their EHR data really entails, and what features are desirable to support that process better. We found that Discovery provided much needed features and could be used in a variety of everyday scenarios, especially for preparing and during clinical visits and also for raising awareness, reflection, and planning. According to the study participants, Discovery provided a robust set of features for supporting independent exploration and sensemaking of their EHR data: summary and quick overview of the data, finding prevalence, periodicity, co-occurrence, and pre-post of medical events, as well as comparing medical record types and subtypes across providers. In addition, we extracted important design implications from the user feedback on data exploration with multiple views and nonstandard user interface elements. Patient-centered sensemaking tools should have a core set of features that can be learned quickly and support common use cases for a variety of users. The patients should be able to detect time-oriented patterns of medical events and get enough context and explanation on demand in a single exploration view that feels warm and familiar and relies on patient-friendly language. However, this view should have enough plasticity to adjust to the patient's information needs as the sensemaking unfolds. Future designs should include the physicians in the patient's sensemaking process and improve the communication in clinical visits and via messaging.

Classification of Right/Left Hand Motor Imagery by Effective Connectivity Based on Transfer Entropy in Electroencephalogram Signal.

The right and left-hand motor imagery (MI) analysis based on the electroencephalogram (EEG) signal can directly link the central nervous system to a computer or a device. This study aims to identify a set of robust and nonlinear effective brain connectivity features quantified by transfer entropy (TE) to characterize the relationship between brain regions from EEG signals and create a hierarchical feature selection and classification for discrimination of right and lefthand MI tasks. TE is calculated among EEG channels as the distinctive, effective connectivity features. TE is a model-free method that can measure nonlinear effective connectivity and analyze multivariate dependent directed information flow among neural EEG channels. Then four feature subset selection methods namely relief-F, Fisher, Laplacian, and local learningbased clustering (LLCFS) algorithms are used to choose the most significant effective connectivity features and reduce redundant information. Finally, support vector machine (SVM) and linear discriminant analysis (LDA) methods are used for classification. Results show that the best performance in 29 healthy subjects and 60 trials is achieved using the TE method via the Relief-F algorithm as feature selection and support vector machine (SVM) classification with 91.02% accuracy. The TE index and a hierarchical feature selection and classification can be useful for the discrimination of right- and left-hand MI tasks from multichannel EEG signals. Effective connectivity features were extracted from electroencephalogram (EEG) to analyze relationships between regions.Four feature selection methods used to select most significant effective features.Support vector machine (SVM) used for discrimination of right and left hand motor imagery (MI) task. In this study, we investigated brain activity using effective connectivity during MI task based on EEG signals. The motor imagery task can accomplish the same goal as motor execution, since they are both activated by the same brain area. Transfer entropy, coherence, and Granger casualty were employed to extract the features. Differential patterns of activity between the left vs. right MI task showed activity around the motor area rather than other areas. In order to reduce redundant information and select the most significant effective connectivity features, four feature subset selection algorithms are used: Relief-F, Fisher, Laplacian, and learning-based clustering feature selection (LLCFS). Then, support vector machine (SVM) and linear discriminant analysis (LDA) are used to classify left and right hand MI task. Comparison of three different connectivity methods showed that TE index had the highest classification accuracy, and could be useful for the discrimination of right and left hand MI tasks from multichannel EEG signals.

Comparative analysis of deep learning models for detecting face mask

The spread of Corona Virus Disease 19 (COVID-19) in Indonesia is still relatively high and has not shown a significant decrease. One of the main reasons is due to the lack of supervision on the implementation of health protocols such as wearing masks in daily activities. Recently, state-of-the-art algorithms were introduced to automate face mask detection. To be more specific, the researchers developed various kinds of architectures for the detection of masks based on computer vision methods. This paper aims to evaluate well-known architectures, namely the ResNet50, VGG11, InceptionV3, EfficientNetB4, and YOLO (You Only Look Once) to recommend the best approach in this specific field. By using the MaskedFace-Net dataset, the experimental results showed that the EfficientNetB4 architecture has better accuracy at 95.77% compared to the YOLOv4 architecture of 93.40%, InceptionV3 of 87.30%, YOLOv3 of 86.35%, ResNet50 of 84.41%, VGG11 of 84.38%, and YOLOv2 of 78.75%, respectively. It should be noted that particularly for YOLO, the model was trained using a collection of MaskedFace-Net images that had been pre-processed and labelled for the task. The model was initially able to train faster with pre-trained weights from the COCO dataset thanks to transfer learning, resulting in a robust set of features expected for face mask detection and classification.

Complexity measures and features for times series classification

Time series classification is a growing problem in different disciplines due to the progressive digitalization of the world. The best state-of-the-art algorithms focus on performance, seeking the best possible results, leaving interpretability at a second level, if any. Furthermore, interpretable proposals are far from providing competitive results. In this work, focused on time series classification, we propose a new representation of time series based on a robust and complete set of features. This new representation allows extracting more meaningful information on the underlying time series structure to develop effective classifiers whose results are much easier to interpret than current state-of-the-art models. The proposed feature set allows using the traditional vector-based classification algorithms in time series problems, significantly increasing the number of techniques available for this type of problem. To evaluate the performance of our proposal, we have used the state-of-the-art repository of time series classification, UCR, composed of 112 datasets. The experimental results show that through this representation, more interpretable classifiers can be obtained which are competitive. More specifically, they obtain no statistically significant differences from the second and third-best models of the state-of-the-art. Apart from competitive results in accuracy, our proposal is able to improve the model interpretability based on the set of features proposed.

An integrated decision support system for heart failure prediction based on feature transformation using grid of stacked autoencoders

Heart failure (HF) is a life-threatening disease. Therefore, prediction of risks associated with HF is essential at the early stages of the disease in order to ensure safety and reduce the mortality rate associated with it. To meet this objective, numerous medical decision support systems (MDSS) have been developed for HF risk prediction. Most of these MDSS focus on feature transformation-based methods and their integration with machine learning models. However, careful investigation of these MDSS reveals that they improved the accuracy on the testing data while the performance on the training data was either compromised or not taken into account. Such an improvement is not pertinent from machine learning perspective. Furthermore, the produced HF prediction accuracy also needs considerable amount of improvement. To tackle such issues, this study proposes a more robust approach that integrates a grid of stacked autoencoders with neural network model. The stacked autoencoders transform features by yielding more robust set of features and the neural network classifies the newly extracted set of features. The strength of the proposed integrated MDSS is evaluated by using an online HF benchmark database namely Cleveland database. The performance of the proposed feature transformation method is also compared with other well known methods namely Principal Component Analysis (PCA), Kernel PCA (KPCA) and feature selection methods i.e., chi-square, Gini index, F-score and mutual information based feature selection. The proposed approach outperformed other state-of-the-art feature transformation methods. The proposed integrated MDSS achieves classification accuracy of 95.55%, specificity of 97.95%, and sensitivity of 92.68%.

Incremental Adaptive Gesture Classifier for Upper Limb Prostheses

Myoelectric pattern recognition is widely used to control upper limb prostheses. However, the non-stationary characteristics of electromyography (EMG) signals, caused by physiological changes (e.g. muscle fatigue) or non-physiological changes (e.g. the electrode- skin impedance), hinder the use of prostheses and deteriorate the performance of the myoelectric control system. In this paper, a set of robust features is proposed to be integrated with adaptive learning techniques in order to improve the myoelectric performance. Four types of features are proposed, namely the H, HI, AI, and AIH features. The H features correspond to the histogram-oriented gradient (HOG) algorithm of the High-Density surface Electromyography map (HD-sEMG map). On the other hand, the HI features are generated by combining the H features and the intensity feature that is evaluated from the HD-sEMG map. AI is the intensity features calculated from the segmented HD-sEMG maps constructed in the individual channel. Finally, AIH features are obtained by combining the H features and AI features. Offline and online adaptive tests are conducted to evaluate the proposed features. The results show that employing the proposed AI features with an adaptive classifier improves the classification accuracy from 91.58% to 97.2% in online classification setups. Results also show that the AI features are more robust against noise than TD features. The average classification accuracy is reduced by 0.7% when additive White Gaussian noise is applied, in comparison to 5.8% reduction in accuracy when TD features is used. The results confirm the robustness of the proposed features extracted from the HD-sEMG map.