Feature-based Model Research Articles

Representational Similarity Analysis for Tracking Neural Correlates of Haptic Learning on a Multimodal Device.

A goal of wearable haptic devices has been to enable haptic communication, where individuals learn to map information typically processed visually or aurally to haptic cues via a process of cross-modal associative learning. Neural correlates have been used to evaluate haptic perception and may provide a more objective approach to assess association performance than more commonly used behavioral measures of performance. In this article, we examine Representational Similarity Analysis (RSA) of electroencephalography (EEG) as a framework to evaluate how the neural representation of multifeatured haptic cues changes with association training. We focus on the first phase of cross-modal associative learning, perception of multimodal cues. A participant learned to map phonemes to multimodal haptic cues, and EEG data were acquired before and after training to create neural representational spaces that were compared to theoretical models. Our perceptual model showed better correlations to the neural representational space before training, while the feature-based model showed better correlations with the post-training data. These results suggest that training may lead to a sharpening of the sensory response to haptic cues. Our results show promise that an EEG-RSA approach can capture a shift in the representational space of cues, as a means to track haptic learning.

Reduced-order hybrid modelling for powder compaction: Predicting density and classifying diametrical hardness

Drawing on machine learning (ML) techniques and physics-based modelling, two feature-based reduced-order models are presented: one for the quantitative prediction of density and another for the classification of the diametrical hardness of pellets from a powder compaction process (pelleting). For interpretability, the models use as input only the parameters from a modified Drucker–Prager Cap (DPC) model calculated from process data monitoring and the applied maximal compression force. For quantitative density prediction, 8 features linked to first-principles models of powder compaction are generated, and the final model uses only 2. A critical part of the modelling, and also one of the main contributions, is a data augmentation step for the primary data set of this study by leveraging much smaller supplementary data sets that have measurements not present in the primary data set.The final results imply a significant reduction in the quantity of data needed for model input and cut down the cost of data acquisition, storage, and computational time. Additionally provided is a detailed analysis of the impact and relevance of the generated features on the model performance.The density prediction model, using only 2 features, reaches a mean absolute scaled error (MASE) of 12.9% and a mean absolute error (MAE) of 0.10 (where r2=0.975). The scaled (diametrical) hardness classifier has an F1 score of 0.915 using 4 features.

Predicting ankle and knee sagittal kinematics and kinetics using an ankle-mounted inertial sensor

The purpose of this study was to develop a machine learning model to reconstruct time series kinematic and kinetic profiles of the ankle and knee joint across six different tasks using an ankle-mounted IMU. Four male collegiate basketball players performed repeated tasks, including walking, jogging, running, sidestep cutting, max-height jumping, and stop-jumping, resulting in a total of 102 movements. Ankle and knee flexion-extension angles and moments were estimated using motion capture and inverse dynamics and considered ‘actual data’ for the purpose of model fitting. Synchronous acceleration and angular velocity data were collected from right ankle-mounted IMUs. A time-series feature extraction model was used to determine a set of features used as input to a random forest regression model to predict the ankle and knee kinematics and kinetics. Five-fold cross-validation was performed to verify the model accuracy, and statistical parametric mapping was used to determine the difference between the predicted and experimental time series. The random forest regression model predicted the time-series profiles of the ankle and knee flexion-extension angles and moments with high accuracy (Kinematics: R2 ranged from 0.782 to 0.962, RMSE ranged from 2.19° to 11.58°; Kinetics: R2 ranged from 0.711 to 0.966, RMSE ranged from 0.10 Nm/kg to 0.41 Nm/kg). There were differences between predicted and actual time series for the knee flexion-extension moment during stop-jumping and walking. An appropriately trained feature-based regression model can predict time series knee and ankle joint angles and moments across a wide range of tasks using a single ankle-mounted IMU.

Circular economy e-business model portfolio development for e-business applications: Impacts on ESG and sustainability performance

The circular economy has emerged as a key driver in the development of sustainability in the current digital world. Circular economy businesses have generated significant contributions in terms of gross domestic product and employment. In the future, new approaches are needed that meet global trends in digitalisation, resources, and ecological challenges. Instead of solely addressing environmental issues, e-businesses in the digital age need to ensure the sustainable development of the circular economy. This paper proposes a feature-based circular economy e-business model for uplifting environment, social, and governance (ESG) and sustainability performance. We have employed indicators that support the implementation of the circular economy process related to technical aspects, reduce the cost of operation and resource pressure, minimise the environmental impact, and maximise social and economic impacts. We present here a comprehensive discussion that provides an in-depth literature review, an emerging e-business model for the circular economy, business, e-business application portfolio, and software application development for sustainable circular economy Business.

A computer-aided feature-based encryption model with concealed access structure for medical Internet of Things

One of the Internet of Things (IoT) security issues is the secure sharing and granular management of data access. This study recommends a feature-based encryption scheme with a hidden access structure for medical IoT data security. While establishing fine-grained access control of ciphertext data, the system can guarantee clinical client data privacy. First, it is recommended to convert identity-based encryption (IBE) into a feature-based encryption model (FBEM) using a universal conversion technique that supports multi-valued attributes and gates. IBE characteristics could be inherited by the converted FBEM. The conversion method is then used to change the receiver anonymous IBE scheme into the FBEM scheme with concealed access structure. The FBEM model is then used to construct the IoT scenario for the smart medical application. Theoretical analysis and experimental findings reveal that the suggested system provides advantages over prominent systems regarding computing efficiency, storage load, and security when the access structure is disguised.

Personalized Machine Learning Approach to Estimating Knee Kinematics Using Only Shank-Mounted IMU

Knee kinematics is a valuable measure of knee joint function. However, collecting that data outside the clinic is difficult, especially with a limited number of wearable sensors and when you only use an ankle-mounted inertial measurement unit to estimate knee kinematics. Due to the cyclic nature of gait, it is possible to use machine learning to extract joint angles from only ankle-mounted sensors. This study aimed to use time-series feature extraction and a random forest regressor to generate a person-specific surrogate model for estimating knee joint flexion angles from a single-mounted inertial measurement unit above the ankle. Optical motion capture and inertial data from 10 healthy participants walking on a treadmill were collected to create 10 personalised surrogate models for estimating right knee flexion angles during gait. An additional 10 models were created for a leave-one-out analysis to test the generalisability of the models. Temporal cross-validation of the personalised models and a leave-one-out analysis was performed on the selected feature set. The personalised models achieved an average root mean square error of 2.45 ± 0.65 degrees (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.98) compared to a gold-standard optical motion capture. The generalised models achieved an average root mean square error of 6.77 ± 3.38 degrees (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.83) in the leave-one-out analysis. Time series feature-based personalised surrogate models could be used to accurately estimate knee kinematics by using a single ankle-mounted sensor. However, more data is required to train a generalised model using the presented method.

Enhanced Semantic Representation Learning for Sarcasm Detection by Integrating Context-Aware Attention and Fusion Network.

Sarcasm is a sophisticated figurative language that is prevalent on social media platforms. Automatic sarcasm detection is significant for understanding the real sentiment tendencies of users. Traditional approaches mostly focus on content features by using lexicon, n-gram, and pragmatic feature-based models. However, these methods ignore the diverse contextual clues that could provide more evidence of the sarcastic nature of sentences. In this work, we propose a Contextual Sarcasm Detection Model (CSDM) by modeling enhanced semantic representations with user profiling and forum topic information, where context-aware attention and a user-forum fusion network are used to obtain diverse representations from distinct aspects. In particular, we employ a Bi-LSTM encoder with context-aware attention to obtain a refined comment representation by capturing sentence composition information and the corresponding context situations. Then, we employ a user-forum fusion network to obtain the comprehensive context representation by capturing the corresponding sarcastic tendencies of the user and the background knowledge about the comments. Our proposed method achieves values of 0.69, 0.70, and 0.83 in terms of accuracy on the Main balanced, Pol balanced and Pol imbalanced datasets, respectively. The experimental results on a large Reddit corpus, SARC, demonstrate that our proposed method achieves a significant performance improvement over state-of-art textual sarcasm detection methods.

An Empirical Comparison of Interpretable Models to Post-Hoc Explanations

Recently, some effort went into explaining intransparent and black-box models, such as deep neural networks or random forests. So-called model-agnostic methods typically approximate the prediction of the intransparent black-box model with an interpretable model, without considering any specifics of the black-box model itself. It is a valid question whether direct learning of interpretable white-box models should not be preferred over post-hoc approximations of intransparent and black-box models. In this paper, we report the results of an empirical study, which compares post-hoc explanations and interpretable models on several datasets for rule-based and feature-based interpretable models. The results seem to underline that often directly learned interpretable models approximate the black-box models at least as well as their post-hoc surrogates, even though the former do not have direct access to the black-box model.

Development of a Gait Feature-Based Model for Classifying Cognitive Disorders Using a Single Wearable Inertial Sensor.

Gait changes are potential markers of cognitive disorders (CDs). We developed a model for classifying older adults with CD from those with normal cognition using gait speed and variability captured from a wearable inertial sensor and compared its diagnostic performance for CD with that of the model using the Mini-Mental State Examination (MMSE). We enrolled community-dwelling older adults with normal gait from the Korean Longitudinal Study on Cognitive Aging and Dementia and measured their gait features using a wearable inertial sensor placed at the center of body mass while they walked on a 14-m long walkway thrice at comfortable paces. We randomly split our entire dataset into the development (80%) and validation (20%) datasets. We developed a model for classifying CD using logistic regression analysis from the development dataset and validated it in the validation dataset. In both datasets, we compared the diagnostic performance of the model with that using the MMSE. We estimated optimal cutoff score of our model using receiver operator characteristic analysis. In total, 595 participants were enrolled, of which 101 of them experienced CD. Our model included both gait speed and temporal gait variability and exhibited good diagnostic performance for classifying CD from normal cognition in both the development (area under the receiver operator characteristic curve [AUC] = 0.788, 95% CI 0.748-0.823, p < 0.001) and validation datasets (AUC = 0.811, 95% CI 0.729-0.877, p < 0.001). Our model showed comparable diagnostic performance for CD with that of the model using the MMSE in both the development (difference in AUC = 0.026, standard error [SE] = 0.043, z statistic = 0.610, p = 0.542) and validation datasets (difference in AUC = 0.070, SE = 0.073, z statistic = 0.956, p = 0.330). The optimal cutoff score of the gait-based model was >-1.56. Our gait-based model using a wearable inertial sensor may be a promising diagnostic marker of CD in older adults. This study provides Class III evidence that gait analysis can accurately distinguish older adults with CDs from healthy controls.

FA-Net: A Fused Feature for Multi-Head Attention Recoding Network for Pear Leaf Nutritional Deficiency Diagnosis with Visual RGB-Image Depth and Shallow Features

Accurate diagnosis of pear tree nutrient deficiency symptoms is vital for the timely adoption of fertilization and treatment. This study proposes a novel method on the fused feature multi-head attention recording network with image depth and shallow feature fusion for diagnosing nutrient deficiency symptoms in pear leaves. First, the shallow features of nutrient-deficient pear leaf images are extracted using manual feature extraction methods, and the depth features are extracted by the deep network model. Second, the shallow features are fused with the depth features using serial fusion. In addition, the fused features are trained using three classification algorithms, F-Net, FC-Net, and FA-Net, proposed in this paper. Finally, we compare the performance of single feature-based and fusion feature-based identification algorithms in the nutrient-deficient pear leaf diagnostic task. The best classification performance is achieved by fusing the depth features output from the ConvNeXt-Base deep network model with shallow features using the proposed FA-Net network, which improved the average accuracy by 15.34 and 10.19 percentage points, respectively, compared with the original ConvNeXt-Base model and the shallow feature-based recognition model. The result can accurately recognize pear leaf deficiency images by providing a theoretical foundation for identifying plant nutrient-deficient leaves.

An Improved Explicit Reference Modeling Methodology for Parametric Design

The use of formal methodologies for parametric modeling has been shown to significantly improve the robustness and reusability of feature-based CAD models, particularly in response to changes, which has implications in the entire product development process. One of these methodologies, explicit reference modeling, has been proven particularly effective in design scenarios involving complex geometry. However, certain design situations may lead to different interpretations of the methodology and the possibility of drastically different modeling alternatives at certain points in the process, some of which can become inefficient and hinder reusability. In this paper, we analyze the explicit reference modeling methodology through the lens of CAD quality and identify the key aspects that may lead to underperformance. We propose a new improved version of the methodology that maximizes model robustness and flexibility, while reducing regeneration time and intrinsic variability. Five modeling aspects are considered: the internal structure of functional geometries, subtractions, combination operations, replication features, and localized modifications. We validate our proposal through an empirical study in which 96 modeling paths for a parametric model were analyzed.

Identification of human phosphoglycerate mutase 1 (PGAM1) inhibitors using hybrid virtual screening approaches.

PGAM1 plays a critical role in cancer cell metabolism through glycolysis and different biosynthesis pathways to promote cancer. It is generally known as a crucial target for treating pancreatic ductal adenocarcinoma, the deadliest known malignancy worldwide. In recent years different studies have been reported that strived to find inhibitory agents to target PGAM1, however, no validated inhibitor has been reported so far, and only a small number of different inhibitors have been reported with limited potency at the molecular level. Our in silico studies aimed to identify potential new PGAM1 inhibitors that could bind at the allosteric sites. At first, shape and feature-based models were generated and optimized by performing receiver operating characteristic (ROC) based enrichment studies. The best query model was then employed for performing shape, color, and electrostatics complementarity-based virtual screening of the ChemDiv database. The top two hundred and thirteen hits with greater than 1.2 TanimotoCombo score were selected and then subjected to structure-based molecular docking studies. The hits yielded better docking scores than reported compounds, were selected for subsequent structural similarity-based clustering analysis to select the best hits from each cluster. Molecular dynamics simulations and binding free energy calculations were performed to validate their plausible binding modes and their binding affinities with the PGAM1 enzyme. The results showed that these compounds were binding in the reported allosteric site of the enzyme and can serve as a good starting point to design better active selective scaffolds against PGAM1enzyme.

Stacked machine learning models to classify atrial disorders based on clinical ECG features: a method to predict early atrial fibrillation.

Atrial Tachycardia (AT) and Left Atrial Enlargement (LAE) are atrial diseases that are significant precursors to Atrial Fibrillation (AF). There are ML models for ECG classification; clinical features-based classification is required. The suggested work aims to create stacked ML models that categorize Sinus Rhythm (SR), Sinus Tachycardia (ST), AT, and LAE signals based on clinical parameters for AF prognosis. The classification was based on thirteen clinical parameters, such as amplitude, time domain ECG aspects, and P-Wave Indices (PWI), such as the ratio of P-wave length and amplitude ((P (ms)/P (µV)), P-wave area (µV*ms), and P-wave terminal force (PTFV1(µV*ms). Apart from classifying the ECG signals, the stacked ML models prioritized the clinical features using a pie formula-based technique. The Stack 1 model achieves 99% accuracy, sensitivity, precision, and F1 score, while the Stack 2 model achieves 91%, 91%, 94%, and 92% for identifying SR, ST, LAE, and AT, respectively. Both stack models obtained a computational time of 0.06 seconds. PTFV1 (µV*ms), P (ms)/P (µV)), and P-wave area (µV*ms) were ranked as crucial clinical features. Clinical feature-based stacking ML models may help doctors obtain insight into important clinical ECG aspects for early AF prediction.

Semantically Adaptive JND Modeling with Object-Wise Feature Characterization, Context Inhibition and Cross-Object Interaction.

Performance bottlenecks in the optimization of JND modeling based on low-level manual visual feature metrics have emerged. High-level semantics bear a considerable impact on perceptual attention and subjective video quality, yet most existing JND models do not adequately account for this impact. This indicates that there is still much room and potential for performance optimization in semantic feature-based JND models. To address this status quo, this paper investigates the response of visual attention induced by heterogeneous semantic features with an eye on three aspects, i.e., object, context, and cross-object, to further improve the efficiency of JND models. On the object side, this paper first focuses on the main semantic features that affect visual attention, including semantic sensitivity, objective area and shape, and central bias. Following that, the coupling role of heterogeneous visual features with HVS perceptual properties are analyzed and quantified. Second, based on the reciprocity of objects and contexts, the contextual complexity is measured to gauge the inhibitory effect of contexts on visual attention. Third, cross-object interactions are dissected using the principle of bias competition, and a semantic attention model is constructed in conjunction with a model of attentional competition. Finally, to build an improved transform domain JND model, a weighting factor is used by fusing the semantic attention model with the basic spatial attention model. Extensive simulation results validate that the proposed JND profile is highly consistent with HVS and highly competitive among state-of-the-art models.

How Many Data Does Machine Learning in Human–Computer Interaction Need?: Re-Estimating the Dataset Size for Convolutional Neural Network-Based Models of Visual Perception

Artificial intelligence (AI)-based user-interface (UI) design and evaluation are currently constrained by the scarcity of human-generated training data. Correspondingly, choosing appropriate neural network (NN) architecture and carefully planning the sample size is essential for building accurate machine learning models. Previously, we have estimated that for a convolutional NN to produce better mean square errors (MSEs) than feature-based models, the required training dataset size should be approximately 3000. Our current validation study with roughly 4000 web UIs and 233 subjects suggests that the estimation should be closer to 17,000. We propose corrected regression models suggesting that the dataset size effect is better described using a logarithmic function. We also report significant differences in MSEs between the employed perception dimensions, with Aesthetics models having an MSE 21.5% worse than Complexity and 12.1% worse than Orderliness.

Enhancer-FRL: Improved and Robust Identification of Enhancers and Their Activities Using Feature Representation Learning.

Enhancers are crucial for precise regulation of gene expression, while enhancer identification and strength prediction are challenging because of their free distribution and tremendous number of similar fractions in the genome. Although several bioinformatics tools have been developed, shortfalls in these models remain, and their performances need further improvement. In the present study, a two-layer predictor called Enhancer-FRL was proposed for identifying enhancers (enhancers or nonenhancers) and their activities (strong and weak). More specifically, to build an efficient model, the feature representation learning scheme was applied to generate a 50D probabilistic vector based on 10 feature encodings and five machine learning algorithms. Subsequently, the multiview probabilistic features were integrated to construct the final prediction model. Compared with the single feature-based model, Enhancer-FRL showed significant performance improvement and model robustness. Performance assessment on the independent test dataset indicated that the proposed model outperformed state-of-the-art available toolkits. The webserver Enhancer-FRL is freely accessible at http://lab.malab.cn/∼wangchao/softwares/Enhancer-FRL/, The code and datasets can be downloaded at the webserver page or at the Github https://github.com/wangchao-malab/Enhancer-FRL/.

Time-Frequency Feature-Based Seismic Response Prediction Neural Network Model for Building Structures

Currently, machine learning techniques are widely used in structural seismic response studies. The developed network models for various types of seismic response provide new ways to analyse seismic hazards. However, it is not easy to balance the applicability of the input, accuracy, and computational efficiency for existing network models. In this paper, a neural network model containing an efficient self-adaptive feature extraction module (AFEM) is designed. It can recognize time-frequency features from ground motion (GM) inputs for structural seismic response prediction tasks while considering the model’s computational accuracy and computational cost. The self-adaptive feature extraction module is constructed based on the MFCCs feature extraction process in NLP. AFEM recognizes time-frequency features closely related to structures’ behaviour and response under dynamic loads. Taking the seismic response prediction of a typical building as the target task, the neural network configuration, including a baseline model M0 and three comparison models (M1, M2, and M3) with AFEM, is systematically analysed. The results demonstrate that the proposed M1 model with initial AFEM, the M2 model with combined amplitude and phase features, and the M3 model with a complex-valued network are more adaptable than the baseline model to the target task. The extracted amplitude and phase features by the M3 model’s AFEM significantly improve model validation accuracy by 8.6% while reducing computation time by 11.4%. It could provide the basis for future research on regional earthquake damage intelligence assessment systems.

Construction of a TTN Mutation-Based Prognostic Model for Evaluating Immune Microenvironment, Cancer Stemness, and Outcomes of Colorectal Cancer Patients.

Colorectal cancer (CRC) is one of the commonest cancers worldwide. As conventional biomarkers cannot clearly define the heterogeneity of CRC, it is essential to establish novel prognostic models. For the training set, data pertaining to mutations, gene expression profiles, and clinical parameters were obtained from the Cancer Genome Atlas. Consensus clustering analysis was used to identify the CRC immune subtypes. CIBERSORT was used to analyze the immune heterogeneity across different CRC subgroups. Least absolute shrinkage and selection operator regression was used to identify the genes for constructing the immune feature-based prognostic model and to determine their coefficients. A gene prognostic model was then constructed to predict patient outcomes; the model was then externally validated using data from the Gene Expression Omnibus. As a high-frequency somatic mutation, the titin (TTN) mutation has been identified as a risk factor for CRC. Our results demonstrated that TTN mutations have the potential to modulate the tumor microenvironment, converting it into the immunosuppressive type. In this study, we identified the immune subtypes of CRC. Based on the identified subtypes, 25 genes were selected for prognostic model construction; a prediction model was also constructed, and its prediction accuracy was tested using the validation dataset. The potential of the model in predicting immunotherapy responsiveness was then explored. TTN-mutant and TTN-wild-type CRC demonstrated different microenvironment features and prognosis. Our model provides a robust immune-related gene prognostic tool and a series of gene signatures for evaluating the immune features, cancer stemness, and prognosis of CRC.

Magnetic resonance imaging based radiomics prediction of Human Papillomavirus infection status and overall survival in oropharyngeal squamous cell carcinoma

ObjectivesHuman papillomavirus- (HPV) positive oropharyngeal squamous cell carcinoma (OPSCC) differs biologically and clinically from HPV-negative OPSCC and has a better prognosis. This study aims to analyze the value of magnetic resonance imaging (MRI)-based radiomics in predicting HPV status in OPSCC and aims to develop a prognostic model in OPSCC including HPV status and MRI-based radiomics. Materials and methodsManual delineation of 249 primary OPSCCs (91 HPV-positive and 159 HPV-negative) on pretreatment native T1-weighted MRIs was performed and used to extract 498 radiomic features per delineation. A logistic regression (LR) and random forest (RF) model were developed using univariate feature selection. Additionally, factor analysis was performed, and the derived factors were combined with clinical data in a predictive model to assess the performance on predicting HPV status. Additionally, factors were combined with clinical parameters in a multivariable survival regression analysis. ResultsBoth feature-based LR and RF models performed with an AUC of 0.79 in prediction of HPV status. Fourteen of the twenty most significant features were similar in both models, mainly concerning tumor sphericity, intensity variation, compactness, and tumor diameter. The model combining clinical data and radiomic factors (AUC = 0.89) outperformed the radiomics-only model in predicting OPSCC HPV status. Overall survival prediction was most accurate using the combination of clinical parameters and radiomic factors (C-index = 0.72). ConclusionPredictive models based on MR-radiomic features were able to predict HPV status with sufficient performance, supporting the role of MRI-based radiomics as potential imaging biomarker. Survival prediction improved by combining clinical features with MRI-based radiomics.

Estimating the Depth of Anesthesia from EEG Signals Based on a Deep Residual Shrinkage Network.

The reliable monitoring of the depth of anesthesia (DoA) is essential to control the anesthesia procedure. Electroencephalography (EEG) has been widely used to estimate DoA since EEG could reflect the effect of anesthetic drugs on the central nervous system (CNS). In this study, we propose that a deep learning model consisting mainly of a deep residual shrinkage network (DRSN) and a 1 × 1 convolution network could estimate DoA in terms of patient state index (PSI) values. First, we preprocessed the four raw channels of EEG signals to remove electrical noise and other physiological signals. The proposed model then takes the preprocessed EEG signals as inputs to predict PSI values. Then we extracted 14 features from the preprocessed EEG signals and implemented three conventional feature-based models as comparisons. A dataset of 18 patients was used to evaluate the models' performances. The results of the five-fold cross-validation show that there is a relatively high similarity between the ground-truth PSI values and the predicted PSI values of our proposed model, which outperforms the conventional models, and further, that the Spearman's rank correlation coefficient is 0.9344. In addition, an ablation experiment was conducted to demonstrate the effectiveness of the soft-thresholding module for EEG-signal processing, and a cross-subject validation was implemented to illustrate the robustness of the proposed method. In summary, the procedure is not merely feasible for estimating DoA by mimicking PSI values but also inspired us to develop a precise DoA-estimation system with more convincing assessments of anesthetization levels.