Proposed Model Research Articles

Identity, Gender, Age, and Emotion Recognition from Speaker Voice with Multi-task Deep Networks for Cognitive Robotics

AbstractThis paper presents a study on the use of multi-task neural networks (MTNs) for voice-based soft biometrics recognition, e.g., gender, age, and emotion, in social robots. MTNs enable efficient analysis of audio signals for various tasks on low-power embedded devices, thus eliminating the need for cloud-based solutions that introduce network latency. However, the strict dataset requirements for training limit the potential of MTNs, which are commonly used to optimize a single reference problem. In this paper, we propose three MTN architectures with varying accuracy-complexity trade-offs for voice-based soft biometrics recognition. In addition, we adopt a learnable voice representation, that allows to adapt the specific cognitive robotics application to the environmental conditions. We evaluate the performance of these models on standard large-scale benchmarks, and our results show that the proposed architectures outperform baseline models for most individual tasks. Furthermore, one of our proposed models achieves state-of-the-art performance on three out of four of the considered benchmarks. The experimental results demonstrate that the proposed MTNs have the potential for being part of effective and efficient voice-based soft biometrics recognition in social robots.

Sensing PEGylated Peptide Conformations Using a Protein Nanopore.

Membrane pores are exploited for the stochastic sensing of various analytes, and here, we use electrical recordings to explore the interaction of PEGylated peptides of different sizes with a protein pore, CymA. This wide-diameter natural pore comprises densely filled charged residues, facilitating electrophoretic binding of polyethylene glycol (PEG) tagged with a nonaarginine peptide. The small PEG 200 peptide conjugates produced monodisperse blockages and exhibited voltage-dependent translocation across the pores. Notably, the larger PEG 1000 and 2000 peptide conjugates yielded heterogeneous blockages, indicating a multitude of PEG conformations hindering their translocation through the pore. Furthermore, a much larger PEG 5000 peptide occludes the pore entrance, resulting in complete closure. The competitive binding of different PEGylated peptides with the same pore produced specific blockage signals reflecting their identity, size, and conformation. Our proposed model of sensing distinct polypeptide conformations corresponds to disordered protein unfolding, suggesting that this pore can find applications in proteomics.

Data-driven Ai- and Bi-soliton of the cylindrical Korteweg-de Vries (cKdV) equation via prior information PINNs (PIPINNs)

Abstract In this paper, by modifying loss function MS E and training area of the physics-informed neural networks (PINNs), we proposed a neural networks models: prior information PINNs (PIPINNs). We demonstrated the advantages of PIPINNs by simulating Ai- and Bi-soliton solutions of the cylindrical Korteweg-de Vries (cKdV) equation. Numerical experiments showed that our proposed model is not only simulate these solitons of cKdV equation, but also significantly improve the simulation capability. Compared with original PINNs, the prediction accuracy of our proposed models are improved by one to three orders of magnitude. Moreover, the accuracy of PIPINNs is improved again by adding the restriction of conservation of energy.

Alpha-EIOU-YOLOv8: An Improved Algorithm for Rice Leaf Disease Detection

Early detection of plant leaf diseases is a major necessity for controlling the spread of infections and enhancing the quality of food crops. Recently, plant disease detection based on deep learning approaches has achieved better performance than current state-of-the-art methods. Hence, this paper utilized a convolutional neural network (CNN) to improve rice leaf disease detection efficiency. We present a modified YOLOv8, which replaces the original Box Loss function by our proposed combination of EIoU loss and α-IoU loss in order to improve the performance of the rice leaf disease detection system. A two-stage approach is proposed to achieve a high accuracy of rice leaf disease identification based on AI (artificial intelligence) algorithms. In the first stage, the images of rice leaf diseases in the field are automatically collected. Afterward, these image data are separated into blast leaf, leaf folder, and brown spot sets, respectively. In the second stage, after training the YOLOv8 model on our proposed image dataset, the trained model is deployed on IoT devices to detect and identify rice leaf diseases. In order to assess the performance of the proposed approach, a comparative study between our proposed method and the methods using YOLOv7 and YOLOv5 is conducted. The experimental results demonstrate that the accuracy of our proposed model in this research has reached up to 89.9% on the dataset of 3175 images with 2608 images for training, 326 images for validation, and 241 images for testing. It demonstrates that our proposed approach achieves a higher accuracy rate than existing approaches.

Federated learning with hybrid differential privacy for secure and reliable cross‐IoT platform knowledge sharing

AbstractThe federated learning has gained prominent attention as a collaborative machine learning method, allowing multiple users to jointly train a shared model without directly exchanging raw data. This research addresses the fundamental challenge of balancing data privacy and utility in distributed learning by introducing an innovative hybrid methodology fusing differential privacy with federated learning (HDP‐FL). Through meticulous experimentation on EMNIST and CIFAR‐10 data sets, this hybrid approach yields substantial advancements, showcasing a noteworthy 4.22% and up to 9.39% enhancement in model accuracy for EMNIST and CIFAR‐10, respectively, compared to conventional federated learning methods. Our adjustments to parameters highlighted how noise impacts privacy, showcasing the effectiveness of our hybrid DP approach in striking a balance between privacy and accuracy. Assessments across diverse FL techniques and client counts emphasized this trade‐off, particularly in non‐IID data settings, where our hybrid method effectively countered accuracy declines. Comparative analyses against standard machine learning and state‐of‐the‐art FL approaches consistently showcased the superiority of our proposed model, achieving impressive accuracies of 96.29% for EMNIST and 82.88% for CIFAR‐10. These insights offer a strategic approach to securely collaborate and share knowledge among IoT devices without compromising data privacy, ensuring efficient and reliable learning mechanisms across decentralized networks.

Conversational recommender based on graph sparsification and multi-hop attention

Conversational recommender systems provide users with item recommendations via interactive dialogues. Existing methods using graph neural networks have been proven to be an adequate representation of the learning framework for knowledge graphs. However, the knowledge graph involved in the dialogue context is vast and noisy, especially the noise graph nodes, which restrict the primary node’s aggregation to neighbor nodes. In addition, although the recurrent neural network can encode the local structure of word sequences in a dialogue context, it may still be challenging to remember long-term dependencies. To tackle these problems, we propose a sparse multi-hop conversational recommender model named SMCR, which accurately identifies important edges through matching items, thus reducing the computational complexity of sparse graphs. Specifically, we design a multi-hop attention network to encode dialogue context, which can quickly encode the long dialogue sequences to capture the long-term dependencies. Furthermore, we utilize a variational auto-encoder to learn topic information for capturing syntactic dependencies. Extensive experiments on the travel dialogue dataset show significant improvements in our proposed model over the state-of-the-art methods in evaluating recommendation and dialogue generation.

Life-cycle decisions and general equilibrium in the heterogeneous-agent OLG economy

ABSTRACT This study develops a life-cycle model to predict an overlapping generations economy with heterogeneous agents. In our proposed model, ex-ante heterogeneity includes the work probability, survival rate, subjective discount rate, and intertemporal elasticity of substitution. We observe differences in financial decisions and asset accumulation among heterogeneous agents and calculate equilibrium values of aggregate economic variables, such as consumption and capital. For the numerical analyses, we classify the population into four groups based on labour productivity and gender, and investigate the life-cycle behaviours of heterogeneous agents. Retirees in the male group with low labour productivity and low remaining life expectancy have the highest propensity to consume and the highest consumption elasticity for retirement shock compared to other retirees. Workers with relatively high labour productivity have lower risk adjustment factors, indicating that they are less concerned about retirement risk than less productive workers.

Waqf for accelerating socioeconomic development: a proposed model with focus on Pakistan

Purpose The unique institution of waqf that was ignored during the colonization of the Muslim areas has to be revived to play its role in shared growth, social inclusion and cohesion in society. This research paper aims to explore the role of waqf as an instrument for a sustainable growth system and to suggest a model for socioeconomic development in an economy like that of Pakistan. Design/methodology/approach This qualitative research is based on analytical methods to arrive at the frameworks and a model that could facilitate the revival of waqf for community development/social inclusion in economies like that of Pakistan. Findings As most of the OIC member states like Pakistan are facing serious financial problems due to debt servicing obligations, promoting Waqf for various socioeconomic and cultural functions is a vital requirement for such economies. The inability of the state institutions in providing necessary civic, health and education facilities to the public is causing serious harm to the balance of the society. It requires promoting a formal system of charity and using FinTech for waqf-based donations and financing the micro businesses. The perpetuity complimented by the profitability of the waqf properties makes the waqf institutions sustainable and effective when compared to individual charities. Research limitations/implications This is conceptual research discussing the potential of waqf in light of its historical role. Researchers may undertake empirical studies on awqaf operations in various jurisdictions and their role in the empowerment of the poor. Practical implications The research will provide the researchers with insight into the potential of waqf as a tool for community development. Besides, it will enable policymakers and implementation authorities to socialize charity for sustained benefits and welfare. Originality/value To the best of the authors’ knowledge, it is the first major research that discusses the role of waqf in economies facing budgetary and trade deficits in the eradication of poverty and the promotion of social and economic entrepreneurship in realizing the community development targets for the economies like that of Pakistan.

A self-supervised learning model based on variational autoencoder for limited-sample mammogram classification

Deep learning models have found extensive application in medical imaging analysis, particularly in mammography classification. However, these models encounter challenges associated with limited annotated mammography public datasets. In recent years, self-supervised learning (SSL) has emerged as a noteworthy solution to addressing data scarcity by leveraging pretext and downstream tasks. Nevertheless, we recognize a notable scarcity of self-supervised learning models designed for the classification task in mammography. In this context, we propose a novel self-supervised learning model for limited-sample mammogram classification. Our proposed SSL model comprises two primary networks. The first is a pretext task network designed to learn discriminative features through mammogram reconstruction using a variational autoencoder (VAE). Subsequently, the downstream network, dedicated to the classification of mammograms, uses the encoded space extracted by the VAE as input through a simple convolutional neural network. The performance of the proposed model is assessed on public INbreast and MIAS datasets. Comparative analyzes are conducted for the proposed model against previous studies for the same classification task and dataset. The proposed SSL model demonstrates high performance with an AUC of 0.94 for density, 0.99 for malignant-nonmalignant classifications on INbreast, 0.97 for benign-malignant, 0.99 for density, and 0.99 for normal-benign-malignant classifications on MIAS. Additionally, the proposed model reduces computational costs with only 228 trainable parameters, 204.95K FLOPs, and a depth of 3 in mammogram classification. Overall, the proposed SSL model exhibits a robust network architecture characterized by repeatability, consistency, generalization ability, and transferability among datasets, providing less computational complexity than previous studies.Graphical abstract

Investigating Factors Associated with the Development of Postnatal Depression After Cesarean Delivery: A Validation Cohort Study.

This study aimed to validate a proposed association model previously published to determine the clinical relevance of pre-operative determinants in the development of PND after Cesarean delivery (CD). Parturients undergoing elective CD under neuraxial anesthesia were recruited for a prospective cohort study between Oct 2021 and Oct 2022 at KK Women's and Children's Hospital, Singapore. Predelivery pain, psychological and mechanical temporal summation, and demographic data were recorded. A follow-up survey was conducted at 6 to 10 weeks after CD. The primary outcome was the incidence of PND, defined as an Edinburgh Postnatal Depression Scale (EPDS) score ≥ 10. A total of 180 patients were recruited for validation. PND 6 to 10 weeks post-delivery occurred in 18.9% of recruited parturients. Multivariate regression analyses showed that higher pre-operative CSI scores (p=0.0156), higher anxiety levels about upcoming surgery (p=0.0429), increased pre-operative pain scores on movement (p=0.0110), and higher pre-operative HADS subscale scores on anxiety (p=0.0041) were independently associated with the development of PND weeks post-CD. Lower anticipation of pain medication needs (p=0.0038) was independently associated with the development of PND post-CD. The area under curve (AUC) of this multivariable model (training cohort), internal cross validation (training cohort) and external cross validation (validation cohort) were 0.818 (95% CI, 0.746 to 0.889), 0.785 (95% CI, 0.707 to 0.864) and 0.604 (95% CI, 0.497 to 0.710) respectively. The proposed model performed well in a local population. Further refinement is necessary to test the proposed model in populations with social and cultural differences.

Finite element solution of crack-tip fields for an elastic porous solid with density-dependent material moduli and preferential stiffness

In this paper, the finite element solutions of crack-tip fields for an elastic porous solid with density-dependent material moduli are presented. Unlike the classical linearized case in which material parameters are globally constant under a small strain regime, the stiffness of the model presented in this paper can depend upon the density with a modeling parameter. The proposed constitutive relationship appears linear in the Cauchy stress and linearized strain independently. From a subclass of the implicit constitutive relation, the governing equation is bestowed via the balance of linear momentum, resulting in a quasi-linear partial differential equation (PDE) system. Using the classical damped Newton’s method, the sequence of linear problems is then obtained, and the linear PDEs are discretized through a bilinear continuous Galerkin-type finite element method. We perform a series of numerical simulations for material bodies with a single edge-crack subject to a variety of loading types (i.e. the pure mode-I, II, and mixed-mode). Numerical solutions demonstrate that the modeling parameter in our proposed model can control preferential mechanical stiffness with its sign and magnitude along with the change of volumetric strain. This study can provide a mathematical and computational foundation to further model the quasi-static and dynamic evolution of cracks, utilizing the density-dependent moduli model and its modeling framework.

A Semi-Supervised Multi-Scale Arbitrary Dilated Convolution Neural Network for Pediatric Sleep Staging.

Sleep staging is essential for assessing sleep quality and diagnosing sleep disorders. However, sleep staging is a labor-intensive process, making it arduous to obtain large quantities of high-quality labeled data for automatic sleep staging. Meanwhile, most of the research on automatic sleep staging pays little attention to pediatric sleep staging. To address these challenges, we propose a semi-supervised multi-scale arbitrary dilated convolution neural network (SMADNet) for pediatric sleep staging using the scalogram with a high height-to-width ratio generated by the continuous wavelet transform (CWT) as input. To extract more extended time dimensional feature representations and adapt to scalograms with a high height-to-width ratio in SMADNet, we introduce a multi-scale arbitrary dilation convolution block (MADBlock) based on our proposed arbitrary dilated convolution (ADConv). Finally, we also utilize semi-supervised learning as the training scheme for our network in order to alleviate the reliance on labeled data. Our proposed model has achieved performance comparable to state-of-the-art supervised learning methods with 30% labels. Our model is tested on a private pediatric dataset and achieved 79% accuracy, 72% kappa, and 75% MF1. Therefore, our model demonstrates a powerful feature extraction capability and has achieved performance comparable to state-of-the-art supervised learning methods with a small number of labels.

Enhanced quantum resources via two distant atom-cavity systems under the influence of atomic dissipation

Quantum resources such as entanglement and coherence are the holy grail for modern quantum technologies. Although the unwanted environmental effects tackle quantum information processing tasks, suprisingly these key quantum resources may be protected and even enhanced by the implementation of some special hybrid open quantum systems. Here, we aim to show how a dissipative atom-cavity-system can be accomplished to generate enhanced quantum resources. To do so, we consider a couple of dissipative cavities, where each one contains two effective two-level atoms interacting with a single-mode cavity field. In practical applications, a classical laser field may be applied to drive each atomic subsystem. After driving the system, a Bell-state measurement is performed on the output of the system to quantify the entanglement and coherence. The obtained results reveal that the remote entanglement and coherence between the atoms existing inside the two distant cavities are not only enhanced, but can be stabilized, even under the action of dissipation. In contrast, the local entanglement between two atoms inside each dissipative cavity attenuates due to the presence of unwanted environmental effects. Nevertheless, the local coherence may show the same behavior as the remote coherence. Besides, the system provides the steady state entanglement in various interaction regimes, particularly in the strong atom-cavity coupling and with relatively large detuning. More interestingly, our numerical analyses demonstrate that the system may show a memory effect due to the fact that the death and revival of the entanglement take place during the interaction. Our proposed model may find potential applications for the implementation of long distance quantum networks. In particular, it facilitates the distribution of quantum resources between the nodes of large-scale quantum networks for secure communication.

A proposed model using Naïve Bayes and generalized linear models for early detection of heart attack risk

<span>Timely identification of diseases, particularly heart attacks is crucial for individuals, particularly the elderly, to accurately anticipate the onset of the disease based on symtoms. The objective of this study is to develop a highly accurate model for early detection of heart disease using the Naïve Bayes (<span>NB) and generalized linear model (GLM) techniques. In addition, another concern is the model’s subfar accuracy levels, promting the implementation of measures to optimize it. The suggested approach fot optimization involves</span> <span>the utilization of a genetic algorithm (GA). The research findings indicate that the NB and GLM approaches achive a reasonably high level of accuracy. Specifically, the NB model achieves an</span> <span>accuracy of 82.53%, while the GLM achieves an accuracy of 84.50%. Following optimization, the accuracy levels notably improved, with the NB_M-GA model reaching 85.83% and the GLM_M-GA model achieving 86,48%.</span></span>

Improved thermal network model for hot spot location in traction transformer windings

Obtaining accurate location information of the hot spot in advance is a prerequisite for locating the hot spot. To tackle this problem, we propose an improved thermal network model, which provides a quantitative method for placing fiber optic sensors at hot-spot locations. The initial step is to construct an enhanced thermal network model that takes into account the thermal resistance of the boundary layer using heat transfer theory. Then, we compare this model with Computational Fluid Dynamics (CFD) in example calculations to validate its effectiveness and precision. The results show that our proposed model can accurately and rapidly calculate temperature distribution within winding areas, enabling the identification of hot spots. This information is reliable for installing monitoring equipment to measure transformer temperatures.

Mitigating type 1 error inflation and power loss in GxE PRS: Genotype-environment interaction in polygenic risk score models.

The use of polygenic risk score (PRS) models has transformed the field of genetics by enabling the prediction of complex traits and diseases based on an individual's genetic profile. However, the impact of genotype-environment interaction (GxE) on the performance and applicability of PRS models remains a crucial aspect to be explored. Currently, existing genotype-environment interaction polygenic risk score (GxE PRS) models are often inappropriately used, which can result in inflated type 1 error rates and compromised results. In this study, we propose novel GxE PRS models that jointly incorporate additive and interaction genetic effects although also including an additional quadratic term for nongenetic covariates, enhancing their robustness against model misspecification. Through extensive simulations, we demonstrate that our proposed models outperform existing models in terms of controlling type 1 error rates and enhancing statistical power. Furthermore, we apply the proposed models to real data, and report significant GxE effects. Specifically, we highlight the impact of our models on both quantitative and binary traits. For quantitative traits, we uncover the GxE modulation of genetic effects on body mass index by alcohol intake frequency. In the case of binary traits, we identify the GxE modulation of genetic effects on hypertension by waist-to-hip ratio. These findings underscore the importance of employing a robust model that effectively controls type 1 error rates, thus preventing the occurrence of spurious GxE signals. To facilitate the implementation of our approach, we have developed an innovative R software package called GxEprs, specifically designed to detect and estimate GxE effects. Overall, our study highlights the importance of accurate GxE modeling and its implications for genetic risk prediction, although providing a practical tool to support further research in this area.

Computer vision based deep learning approach for toxic and harmful substances detection in fruits

Formaldehyde (CH₂O) is one of the significant chemicals mixed with different perishable fruits in Bangladesh. The fruits are artificially preserved for extended periods by dishonest vendors using this dangerous chemical. Such substances are complicated to detect in appearance. Hence, a reliable and robust detection technique is required. To overcome this challenge and address the issue, we introduce comprehensive deep learning-based techniques for detecting toxic substances. Four different types of fruits, both in fresh and chemically mixed conditions, are used in this experiment. We have applied diverse data augmentation techniques to enlarge the dataset. The performance of four different pre-trained deep learning models was then assessed, and a brand-new model named “DurbeenNet,” created especially for this task, was presented. The primary objective was to gauge the efficacy of our proposed model compared to well-established deep learning architectures. Our assessment centered on the models' accuracy in detecting toxic substances. According to our research, GoogleNet detected toxic substances with an accuracy rate of 85.53 %, VGG-16 with an accuracy rate of 87.44 %, DenseNet with an impressive accuracy rate of 90.37 %, and ResNet50 with an accuracy rate of 91.66 %. Notably, the proposed model, DurbeenNet, outshone all other models, boasting an impressive accuracy rate of 96.71 % in detecting toxic substances among the sample fruits.

An intelligent neural network model to detect red blood cells for various blood structure classification in microscopic medical images

Biomedical image analysis plays a crucial role in enabling high-performing imaging and various clinical applications. For the proper diagnosis of blood diseases related to red blood cells, red blood cells must be accurately identified and categorized. Manual analysis is time-consuming and prone to mistakes. Analyzing multi-label samples, which contain clusters of cells, is challenging due to difficulties in separating individual cells, such as touching or overlapping cells. High-performance biomedical imaging and several medical applications are made possible by advanced biosensors. We develop an intelligent neural network model that can automatically identify and categorize red blood cells from microscopic medical images using region-based convolutional neural networks (RCNN) and cutting-edge biosensors. Our model successfully navigates obstacles like touching or overlapping cells and accurately recognizes various blood structures. Additionally, we utilized data augmentation as a pre-processing method on microscopic images to enhance the model's computational efficiency and expand the sample size. To refine the data and eliminate noise from the dataset, we utilized the Radial Gradient Index filtering algorithm for imaging data equalization. We exhibit improved detection accuracy and a reduced model loss rate when using medical imagery datasets to apply our proposed model in comparison to existing ResNet and GoogleNet models. Our model precisely detected red blood cells in a collection of medical images with 99% training accuracy and 91.21% testing accuracy. Our proposed model outperformed earlier models like ResNet-50 and GoogleNet by 10-15%. Our results demonstrated that Artificial intelligence (AI)-assisted automated red blood cell detection has the potential to revolutionize and speed up blood cell analysis, minimizing human error and enabling early illness diagnosis.

Scalar invariant transform based deep learning framework for detecting heart failures using ECG signals

Heart diseases are leading to death across the globe. Exact detection and treatment for heart disease in its early stages could potentially save lives. Electrocardiogram (ECG) is one of the tests that take measures of heartbeat fluctuations. The deviation in the signals from the normal sinus rhythm and different variations can help detect various heart conditions. This paper presents a novel approach to cardiac disease detection using an automated Convolutional Neural Network (CNN) system. Leveraging the Scale-Invariant Feature Transform (SIFT) for unique ECG signal image feature extraction, our model classifies signals into three categories: Arrhythmia (ARR), Congestive Heart Failure (CHF), and Normal Sinus Rhythm (NSR). The proposed model has been evaluated using 96 Arrhythmia, 30 CHF, and 36 NSR ECG signals, resulting in a total of 162 images for classification. Our proposed model achieved 99.78% accuracy and an F1 score of 99.78%, which is among one of the highest in the models which were recorded to date with this dataset. Along with the SIFT, we also used HOG and SURF techniques individually and applied the CNN model which achieved 99.45% and 78% accuracy respectively which proved that the SIFT–CNN model is a well-trained and performed model. Notably, our approach introduces significant novelty by combining SIFT with a custom CNN model, enhancing classification accuracy and offering a fresh perspective on cardiac arrhythmia detection. This SIFT–CNN model performed exceptionally well and better than all existing models which are used to classify heart diseases.

Grid-free algorithms for direction-of-arrival trajectory localization.

Direction-of-arrival (DOA) estimation algorithms are crucial in localizing acoustic sources. Traditional localization methods rely on block-level processing to extract the directional information from multiple measurements processed together. However, these methods assume that DOA remains constant throughout the block, which may not be true in practical scenarios. Also, the performance of localization methods is limited when the true parameters do not lie on the parameter search grid. In this paper, two trajectory models are proposed, namely the polynomial and harmonic trajectory models, to capture the DOA dynamics. To estimate trajectory parameters, two gridless algorithms are adopted: (i) Sliding Frank-Wolfe (SFW), which solves the Beurling LASSO problem, and (ii) Newtonized orthogonal matching pursuit (NOMP), which is improved over orthogonal matching pursuit (OMP) using cyclic refinement. Furthermore, our analysis is extended to include multi-frequency processing. The proposed models and algorithms are validated using both simulated and real-world data. The results indicate that the proposed trajectory localization algorithms exhibit improved performance compared to grid-based methods in terms of resolution, robustness to noise, and computational efficiency.