Public Datasets Research Articles

DHKFN

Knowledge tracing is effective in modeling learners' knowledge levels to predict future answering situations based on their past learning history and interaction processes. However, current methods often overlook the impact of knowledge point correlations on answer prediction. This paper design a model based on deep hierarchical knowledge fusion networks (DHKFN). It utilizes statistical approaches to construct correlation matrices to capture the correlations between knowledge points. Subsequently, it constructs a topology graph structure of questions and knowledge points and utilizes a multi-head attention network to learn student interaction information from multiple perspectives, effectively constructing adjacency matrices. Finally, it uses GCN to learn the interaction information representation between deep-level knowledge points from dynamically constructed a topology graph structures. Experimental results on three large public datasets show that DHKFN effectively considers the influence of correlations between different knowledge points, thus showing promising effectiveness.

JobFormer: Skill-Aware Job Recommendation with Semantic-Enhanced Transformer

Job recommendation aims to provide potential talents with suitable job descriptions (JDs) consistent with their career trajectory, which plays an essential role in proactive talent recruitment. In real-world management scenarios, the available JD-user records always consist of JDs, user profiles, and click data, in which the user profiles are typically summarized as the user's skill distribution for privacy reasons. Although existing sophisticated recommendation methods can be directly employed, effective recommendation still has challenges considering the information deficit of JD itself and the natural heterogeneous gap between JD and user profile. To address these challenges, we proposed a novel skill-aware recommendation model based on the designed semantic-enhanced Transformer to parse JDs and complete personalized job recommendation. Specifically, we first model the relative items of each JD and then adopt an encoder with the local-global attention mechanism to better mine the intra-job and inter-job dependencies from JD tuples. Moreover, we adopt a two-stage learning strategy for skill-aware recommendation, in which we utilize the skill distribution to guide JD representation learning in the recall stage, and then combine the user profiles for final prediction in the ranking stage. Consequently, we can embed rich contextual semantic representations for learning JDs, while skill-aware recommendation provides effective JD-user joint representation for click-through rate (CTR) prediction. To validate the superior performance of our method for job recommendation, we present a thorough empirical analysis of large-scale real-world and public datasets to demonstrate its effectiveness and interpretability.

Leveraging Advanced Machine Learning Algorithms for Enhanced Cyberattack Detection on U.S. Business Networks

Cyberattacks' rising volume and sophistication have made conventional security measures, such as firewalls, signature-based intrusion detection systems, and antivirus software, increasingly inadequate. The upsurge of cyber threats has been one of the most pressing predicaments for U.S. organizations in the digital age. With the increased dependence on internet-based forums, cloud computing, and interconnected networks, companies face an advancing number of extreme cyberattacks. The chief objective of this research project is to design and deploy proven machine learning methods to enhance the detection and combating of cyberattacks on U.S. organization networks. This research project retrieved a cyber-attack dataset from Kaggle.com, which had a collection of public datasets of cyber threats. This dataset was curated precisely, offering a realistic representation of cyber-attack scenarios, making it an ideal playground for various analytical tasks. The collection was classified as per the source of the relevant information, such as host-based datasets, network traffic datasets, malware or fraud reports, or a special section for datasets that can be classified according to a specific source. The dataset comprised numerous network traffic attributes such as source and destination IP addresses, ports, protocol, payload size, and attack labels. For this research project, three machine learning algorithms were used, namely: Logistic Regression, XG-Boost and Random Forest. This research project applied performance metrics such as accuracy, precision, recall, and F1 score for the performance of the classification models were considered. The result illustrated that the random forest model was far superior in accuracy compared to the logistic regression model; particularly, it had excellent accuracy. Through the use of advanced machine learning models, organizations will be in a position to devise more dynamic and intelligent security systems that evolve with the threat landscape. These intelligent systems monitor every kind of anomaly, malicious activity, and threat response with unparalleled effectiveness. The findings of this research project have significant implications for enhancing cybersecurity in U.S. organizational networks.

TNF inhibitors target a mevalonate metabolite/TRPM2/calcium signaling axis in neutrophils to dampen vasculitis in Behçet’s disease

TNF inhibitors have been used to treat autoimmune and autoinflammatory diseases. Here we report an unexpected mechanism underlying the therapeutic effects of TNF inhibitors in Behçet’s disease (BD), an autoimmune inflammatory disorder. Using serum metabolomics and peripheral immunocyte transcriptomics, we find that polymorphonuclear neutrophil (PMN) from patients with BD (BD-PMN) has dysregulated mevalonate pathway and subsequently increased farnesyl pyrophosphate (FPP) levels. Mechanistically, FPP induces TRPM2-calcium signaling for neutrophil extracellular trap (NET) and proinflammatory cytokine productions, leading to vascular endothelial inflammation and damage. TNF, but not IL-1β, IL-6, IL-18, or IFN-γ, upregulates TRPM2 expression on BD-PMN, while TNF inhibitors have opposite effects. Results from mice with PMN-specific FPP synthetase or TRPM2 deficiency show reduced experimental vasculitis. Meanwhile, analyses of public datasets correlate increased TRPM2 expressions with the clinical benefits of TNF inhibitors. Our results thus implicate FPP-TRPM2-TNF/NETs feedback loops for inflammation aggravation, and novel insights for TNF inhibitor therapies on BD.

A rolling bearing fault signal denoising algorithm that combines a new adaptive information entropy with a new wavelet threshold function

Abstract Mechanical fault diagnosis is of great significance to industrial automation, and extracting vibration fault signals is one of the important tasks in mechanical health monitoring and fault diagnosis. However, due to the complex working environment of rolling bearings, a large amount of noise makes it difficult to extract vibration fault signals. Denoising the vibration signal of bearings can remove interference noise, simplify the early identification of signal features, and thus improve diagnostic accuracy and mechanical maintenance efficiency. This paper proposes a rolling bearing fault signal denoising algorithm, which constructs a new feature extraction and denoising function. This method first decomposes the noisy signal into Intrinsic Mode Functions (IMFs) by Intrinsic Computing Expressive Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN). Secondly, a new adaptive information entropy threshold function is constructed to extract the noisy IMFs from it. Then, the noisy IMF signal is denoised by a new wavelet threshold function. Finally, the noise-free IMFs are reconstructed to reconstruct the denoised signal. To verify the actual performance of the algorithm, comparative experiments were conducted on a self-collected dataset and a public dataset, the results show that this method improves the continuity of signal reconstruction and can remove various types of vibration fault noise signals more effectively and accurately, \hl{thereby improving the fault detection accuracy by 2%-9%.

D3L-SLAM: A Comprehensive Hybrid Simultaneous Location and Mapping System with Deep Keypoint, Deep Depth, Deep Pose, and Line Detection

Robust localization and mapping are crucial for autonomous systems, but traditional handcrafted feature-based visual SLAM often struggles in challenging, textureless environments. Additionally, monocular SLAM lacks scale-aware depth perception, making accurate scene scale estimation difficult. To address these issues, we propose D3L-SLAM, a novel monocular SLAM system that integrates deep keypoints, deep depth estimates, deep pose priors, and a line detector. By leveraging deep keypoints, which are more resilient to lighting variations, our system improves the robustness of visual SLAM. We further enhance perception in low-texture areas by incorporating line features in the front-end and mitigate scale degradation with learned depth estimates. Additionally, point-line feature constraints optimize pose estimation and mapping through a tightly coupled point-line bundle adjustment (BA). The learned pose estimates refine the feature matching process during tracking, leading to more accurate localization and mapping. Experimental results on public and self-collected datasets show that D3L-SLAM significantly outperforms both traditional and learning-based visual SLAM methods in localization accuracy.

GLE-net: global-local information enhancement for semantic segmentation of remote sensing images

Remote sensing (RS) images contain a wealth of information with expansive potential for applications in image segmentation. However, Convolutional Neural Networks (CNN) face challenges in fully harnessing the global contextual information. Leveraging the formidable capabilities of global information modeling with Swin-Transformer, a novel RS images segmentation model with CNN (GLE-Net) was introduced. This integration gives rise to a revamped encoder structure. The subbranch initiates the process by extracting features at varying scales within the RS images using the Multiscale Feature Fusion Module (MFM), acquiring rich semantic information, discerning localized finer features, and adeptly handling occlusions. Subsequently, Feature Compression Module (FCM) is introduced in main branch to downsize the feature map, effectively reducing information loss while preserving finer details, enhancing segmentation accuracy for smaller targets. Finally, we integrate local features and global features through Spatial Information Enhancement Module (SIEM) for comprehensive feature modeling, augmenting the segmentation capabilities of model. We performed experiments on public datasets provided by ISPRS, yielding notably remarkable experimental outcomes. This underscores the substantial potential of our model in the realm of RS image segmentation within the context of scientific research.

MCI net: mamba- convolutional lightweight self-attention medical image segmentation network.

With the development of deep learning in the field of medical image segmentation, various network segmentation models have been developed. Currently, the most common network models in medical image segmentation can be roughly categorized into pure convolutional networks, Transformer-based networks, and networks combining convolution and Transformer architectures. However, when dealing with complex variations and irregular shapes in medical images, existing networks face issues such as incomplete information extraction, large model parameter sizes, high computational complexity, and long processing times. In contrast, models with lower parameter counts and complexity can efficiently, quickly, and accurately identify lesion areas, significantly reducing diagnosis time and providing valuable time for subsequent treatments. Therefore, this paper proposes a lightweight network named MCI-Net, with only 5.48M parameters, a computational complexity of 4.41, and a time complexity of just 0.263. By performing linear modeling on sequences, MCI-Net permanently marks effective features and filters out irrelevant information. It efficiently captures local-global information with a small number of channels, reduces the number of parameters, and utilizes attention calculations with exchange value mapping. This achieves model lightweighting and enables thorough interaction of local-global information within the computation, establishing an overall semantic relationship of local-global information. To verify the effectiveness of the MCI-Net network, we conducted comparative experiments with other advanced representative networks on five public datasets: X-ray, Lung, ISIC-2016, ISIC-2018, and capsule endoscopy and gastrointestinal segmentation. We also performed ablation experiments on the first four datasets. The experimental results outperformed the other compared networks, confirming the effectiveness of MCI-Net. This research provides a valuable reference for achieving lightweight, accurate, and high-performance medical image segmentation network models.

Global induced local network for infrared: dim small target detection

Abstract It is challenging to detect infrared dim targets submerged in complicated backgrounds due to their small size and faint intensity. The previous attention-based detection networks frequently require global long-range dependence. Significant calculations are required to determine the target's sparse but meaningful position. To prevent wasting calculations on the background, this paper offers a detection network guided by global context for local feature learning, named GILNet (Global Induced Local Network). It designs a global location module (GLM) and a local feature interaction module (LFIM) to capture the global position and features of targets, respectively. More specifically, using global context interaction, the GLM finds the region that might contain dim small targets, that is, the coarse location. In the coarsely located regions, the LFIM further acquires feature information about targets. Next, we also design an eight-directional attention operation to obtain the contour information of targets in the low feature map. It is fused with the high feature map in the multi-directional feature fusion module (MFFM), which retains more semantic and spatial information about targets. Finally, quantitative and qualitative analysis show that the GILNet performs better than eight comparison methods on two public datasets.

An optimized method for dose-effect prediction of traditional Chinese medicine based on 1D-ResCNN-PLS.

We introduce a one-dimensional (1D) residual convolutional neural network with Partial Least Squares (1D-ResCNN-PLS) to solve the covariance and nonlinearity problems in traditional Chinese medicine dose-effect relationship data. The model combines a 1D convolutional layer with a residual block to extract nonlinear features and employs PLS for prediction. Tested on the Ma Xing Shi Gan Decoction datasets, the model significantly outperformed conventional models, achieving high accuracies, sensitivities, specificities, and AUC values, with considerable reductions in mean square error. Our results confirm its effectiveness in nonlinear data processing and demonstrate potential for broader application across public datasets.

A highly naturalistic facial expression generation method with embedded vein features based on diffusion model

Abstract Facial expression generation technology has achieved notable progress in computer vision and artificial intelligence. However, challenges persist regarding background consistency, expression clarity, and detailed representation. Additionally, the instability of generative adversarial networks (GANs) during training affects both image quality and diversity. While diffusion models have demonstrated potential advantages over GANs, research on controllable expression generation remains limited. To address these challenges, this paper proposes a highly natural facial expression generation method based on denoising diffusion implicit models (DDIM) with embedded vein features. This approach avoids adversarial training by employing gradual diffusion to generate specific expressions, thereby enhancing both the diversity and authenticity of the images. Vein features are introduced and embedded within the generated expression images to protect the intellectual property of algorithm-generated digital resources. Firstly, image and expression text guide words are combined as conditional inputs to improve the authenticity and diversity of the generated images. Secondly, a classification coding network is introduced to guide expression generation, thus enhancing the accuracy and consistency of the produced expressions. Furthermore, this paper proposes a vein feature fusion method based on multi-directional local dynamic feature coding operator (MLDFO) and integrates DDIM with frequency-domain watermarking technology to achieve image intellectual property protection. Experimental results demonstrate the effectiveness of this method across several public datasets, including FFHQ, CelebA, FV-USM, and SDUMLA-HMT. Notably, in the CelebA dataset, the average expression recognition rate increased by 11.41%, with a 100.00% recognition rate for happy expressions. The generated expression images exhibit a high degree of authenticity and consistency, and the video conversion tests reveal a natural and smooth effect. These results confirm that this method not only advances facial expression generation technology but also significantly enhances the steganographic protection of images.

SPFDNet: Water Extraction Method Based on Spatial Partition and Feature Decoupling

Extracting water information from remote-sensing images is of great research significance for applications such as water resource protection and flood monitoring. Current water extraction methods aggregated richer multi-level features to enhance the output results. In fact, there is a difference in the requirements for the water body and the water boundary. Indiscriminate multi-feature fusion can lead to perturbation and competition of information between these two types of features during the optimization. Consequently, models cannot accurately locate the internal vacancies within the water body with the external boundary. Therefore, this paper proposes a water feature extraction network with spatial partitioning and feature decoupling. To ensure that the water features are extracted with deep semantic features and stable spatial information before decoupling, we first design a chunked multi-scale feature aggregation module (CMFAM) to construct a context path for obtaining deep semantic information. Then, an information interaction module (IIM) is designed to exchange information between two spatial paths with two fixed resolution intervals and the two paths through. During decoding, a feature decoupling module (FDM) is developed to utilize internal flow prediction to acquire the main body features, and erasing techniques are employed to obtain boundary features. Therefore, the deep features of the water body and the detailed boundary information are supplemented, strengthening the decoupled body and boundary features. Furthermore, the integrated expansion recoupling module (IERM) module is designed for the recoupling stage. The IERM expands the water body and boundary features using expansion and adaptively compensates the transition region between the water body and boundary through information guidance. Finally, multi-level constraints are combined to realize the supervision of the decoupled features. Thus, the water body and boundaries can be extracted more accurately. A comparative validation analysis is conducted on the public datasets, including the gaofen image dataset (GID) and the gaofen2020 challenge dataset (GF2020). By comparing with seven SOTAs, the results show that the proposed method achieves the best results, with IOUs of 91.22 and 78.93, especially in the localization of water bodies and boundaries. By applying the proposed method in different scenarios, the results show the stable capability of the proposed method for extracting water with various shapes and areas.

LIVS: a tightly coupled LiDAR, IMU, and camera-based SLAM system under scene degradation

In increasingly complex environments, single-sensor-based SLAM systems face scene degradation problems such as strong light, dynamic obstacle interference, and lack of structural features in the environment, which severely constrain the localization accuracy and mapping effectiveness of SLAM systems. To address this problem, this paper proposes a tightly coupled laser-IMU-visual SLAM system (LIVS) based on factor graph optimization. LIVS consists of two subsystems: the laser inertial system (LIS) and the visual inertial system (VIS). The pose estimation of the LIS can provide an initialization for the VIS; LiDAR suffers scene degradation in environments lacking structural features, while the VIS provides better constraints for Lidar in this environment. Firstly, a method of selecting keyframes based on ORB features and the relative motion of the mobile robot is proposed as a means of improving the positioning accuracy and tracking ability of the system in the case of strenuous motion, and a strategy of local sliding window optimization is adopted to effectively improve the real-time performance and robustness of the system. Meanwhile, in order to further improve the global map consistency and positioning accuracy, a hybrid closed-loop detection method based on laser and vision is proposed, which utilizes the geometrical features of laser and the image features of vision to accomplish the hybrid closed-loop detection. Secondly, the laser inertia factor, IMU pre-integration factor, vision inertia factor, and closed-loop factor are globally optimized in a factor map framework to achieve accurate estimation of the robot pose. Finally, this paper presents experimental evaluations on both public and homemade datasets and compares the performance of the proposed algorithm with that of other algorithms. Experimental results show that the algorithm in this paper exhibits better performance in both localization accuracy and mapping effectiveness.

VSegNet - A Variant SegNet for Improving Segmentation Accuracy in Medical Images with Class Imbalance and Limited Data

Deep learning methods for many medical image segmentation task encounter challenges like smaller datasets and class imbalance. This study proposes a variant SegNet (vSegNet) designed to deliver significantly accurate and reliable segmentation results on such datasets. The novelty lies in designing encoder and decoder blocks with an appropriate number of convolution layers and using the Dice score and Hausdorff distance (HD) as compound loss function in learning. This study used public datasets consisting of chest X-rays, axial CT slices, foot ulcer images, and subset of SPIDER dataset to benchmark segmentation task of the proposed neural network model with other popular networks like U-Net, SegNet, DeepLabv3+, VGG16, MobileNetV2, and fully convolutional network (FCN). For the segmentation of lungs in chest X-rays, vertebral body in CT, augmented data for the previous case, foot ulcer dataset, and segmentation of vertebrae, intervertebral disks and spinal canal in SPIDER dataset (MRI dataset) respectively, the proposed vSegNet performed with a Dice score of 0.96 ± 0.01, 0.90 ± 0.20, 0.95 ± 0.02, 0.86 ± 0.07, and 0.95 ± 0.01 and the HD of 14.33 ± 7.74, 8.45 ± 7.08, 7.99 ± 6.05, 29.32 ± 25.64, and 8.45 ± 2.81 with respect to the ground truth on the test dataset. These results highlight the effectiveness of the proposed model in delivering both higher segmentation accuracy and improved boundary delineation. The proposed network, vSegNet has demonstrated as an effective model for semantic segmentation on class-imbalanced smaller datasets, surpassing all other networks considered in this study in terms of mIoU, BF score, Dice score, HD, accuracy, precision, recall and F1 score on a variety of anatomical regions and medical imaging modalities.

UeLP: accurate user linkage across social platforms against location errors

Abstract User linkage across social platforms can connect the accounts of the same user across different social networks, which is crucial for the identification of users’ multiple social identities and cross-platform association analysis. Cross-platform user linkage based on location is a typical method in current research. These methods typically rely on check-in data to calculate user similarity. However, different from check-in location, the location data obtained from instant messaging social platforms may contain random errors, leading to low accuracy of user linkage of such methods. To solve this problem, this paper proposes an accurate user linkage method across social platforms against location errors. First, unlike existing methods that employ fixed-size grids, this paper uses a multi-grained spatio-temporal grid to organize data, in order to accurately extract user features from error locations. Then, by extracting coarse-grained movement pattern features from user trajectories, candidate users are filtered out, and a small subset of candidate uses is generated to effectively reduce the search space. Next, we establish a weight model based on grid contribution and motion sequence similarity to extract location and temporal features with stronger user orientation. Finally, according to the weight model, the weighted cluster center distance of trajectories is used to calculate the similarity between two different user trajectories. The user with the highest similarity is selected from the candidate subset to complete the user linkage. The extensive experiments are conducted on six public datasets containing 115 866 trajectories and a self-built dataset with 5358 trajectories. The results show the following: compared with the four existing typical location-based methods $k$-BCT, GS, TF-IDF, and TF-IWF, the accuracy Acc@1 is improved by an average of 33%, 44.94%, 15.2%, and 14.55%, respectively, and the accuracy Acc@3 is improved by 30.52%, 34.67%, 13.84%, and 13.19%, respectively.

A novel image classification framework based on variational quantum algorithms

Image classification is a crucial task in machine learning with widespread practical applications. The existing classical framework for image classification typically utilizes a global pooling operation at the end of the network to reduce computational complexity and mitigate overfitting. However, this operation often results in a significant loss of information, which can affect the performance of classification models. To overcome this limitation, we introduce a novel image classification framework that leverages variational quantum algorithms (VQAs) hybrid approaches combining quantum and classical computing paradigms within quantum machine learning. The major advantage of our framework is the elimination of the need for the global pooling operation at the end of the network. In this way, our approach preserves more discriminative features and fine-grained details in the images, which enhances classification performance. Additionally, employing VQAs enables our framework to have fewer parameters than the classical framework, even in the absence of global pooling, which makes it more advantageous in preventing overfitting. We apply our method to different state-of-the-art image classification models and demonstrate the superiority of the proposed quantum architecture over its classical counterpart through a series of state vector simulation experiments on public datasets. Our experiments show that the proposed quantum framework achieves up to a 9.21% increase in accuracy and up to a 15.79% improvement in F1 score, compared to the classical framework. Additionally, we explore the impact of shot noise on our method through shot-based simulation and find that increasing the number of measurements does not always lead to better results. Selecting an appropriate number of measurements can yield optimal results, even surpassing those obtained from state vector simulation.

HRTBDA: a network for post-disaster building damage assessment based on remote sensing images

ABSTRACT Efficient building damage assessment after disasters is vital for emergency response and loss evaluation, but the task is complicated by diverse building structures and complex environments. Traditional methods using Convolutional Neural Networks (CNNs) struggle to capture global contextual features, limiting damage categorization accuracy. To address this, we introduce the High-Resolution Transformer Architecture for Building Damage Assessment (HRTBDA), which enhances multi-scale feature extraction. A Cross-Attention-Based Spatial Fusion (CSF) module is proposed to utilize the attention mechanism, improving the model’s ability to identify detailed associations in damaged buildings. Additionally, we propose a deep convolution network matching optimization strategy that integrates a multilayer perceptron and expands the receptive field, enhancing global feature perception. HRTBDA’s performance was evaluated on two public datasets and compared with five recent frameworks. The model achieved an F1-score of 86.0% in building localization and 78.4% in damage assessment, with a 4.8% improvement in detecting minor damages. These results demonstrate HRTBDA’s potential for improving building damage assessment and highlight its significant advancements over existing methods.

Dynamic Spatial-Temporal Embedding via Neural Conditional Random Field for Multivariate Time Series Forecasting

How to capture dynamic spatial-temporal dependencies remains an open question in multivariate time series (MTS) forecasting. Although recent advanced spatial-temporal graph neural networks (STGNNs) achieve superior forecasting performance, they either consider pre-defined spatial correlations or simply learn static graphs. Some research has tried to learn many adjacent matrices to reveal time-varying spatial correlations, but they generate discrete graphs which cannot encode evolutionary information and also face computational complexity problem. In this article, we propose two significant plugins to help automatically learn enhanced dynamic spatial-temporal embedding of MTS data: (1) a novel neural conditional random field (CRF) layer. We find that the implicit time-varying spatial dependencies are reflected by the explicit changeable links between edges, and we propose the neural CRF to encode such pairwise changeable evolutionary inter-dependencies; (2) a structure adaptive graph convolution (SAGC) that does not require pre-defined graphs to capture semantically richer spatial correlations. Then, we integrate the neural CRF, SAGC with recurrent neural network to develop a new STGNN paradigm termed Adaptive Spatial-Temporal graph neural network with Conditional Random Field (ASTCRF), which can be trained in an end-to-end fashion. We validate the effectiveness, efficiency, and scalability of ASTCRF on five public benchmark MTS datasets.

Millimeter-Wave Radar Clutter Suppression Based on Cycle-Consistency Generative Adversarial Network

Vehicle-mounted millimeter-wave radar is widely used in autonomous driving systems for its ability to observe road scenes at all times and in all weathers. However, the data collected by millimeter-wave radar are seriously affected by the existence of clutter. This clutter will result in false detection during object detection. To address this issue, a feature extraction network with clutter suppression is necessary. This paper proposes a new clutter suppression method for millimeter-wave Range–Angle (RA) images based on a cycle-consistency generative adversarial network (CycleGAN). The generator of the method can be used as the feature extraction network of the object detection. The method aims to convert cluttered images into clutter-free images by unsupervised learning. In this method, an attention gate (AG) is introduced into the generator, a spatial attention mechanism that improves the ability of the model to automatically learn to focus on the features of targets and suppress the clutter of the background. Additionally, the target consistency loss term is added to the loss function to maintain target integrity while suppressing network training overfitting. The public dataset CRUW is utilized to evaluate the performance of the proposed method, which is compared and analyzed with traditional methods and deep learning methods. Experimental results show that the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) of the proposed method reach 39.846 and 0.990, respectively.

Integrated ensemble CNN and explainable AI for COVID-19 diagnosis from CT scan and X-ray images

In light of the ongoing battle against COVID-19, while the pandemic may eventually subside, sporadic cases may still emerge, underscoring the need for accurate detection from radiological images. However, the limited explainability of current deep learning models restricts clinician acceptance. To address this issue, our research integrates multiple CNN models with explainable AI techniques, ensuring model interpretability before ensemble construction. Our approach enhances both accuracy and interpretability by evaluating advanced CNN models on the largest publicly available X-ray dataset, COVIDx CXR-3, which includes 29,986 images, and the CT scan dataset for SARS-CoV-2 from Kaggle, which includes a total of 2,482 images. We also employed additional public datasets for cross-dataset evaluation, ensuring a thorough assessment of model performance across various imaging conditions. By leveraging methods including LIME, SHAP, Grad-CAM, and Grad-CAM++, we provide transparent insights into model decisions. Our ensemble model, which includes DenseNet169, ResNet50, and VGG16, demonstrates strong performance. For the X-ray image dataset, sensitivity, specificity, accuracy, F1-score, and AUC are recorded at 99.00%, 99.00%, 99.00%, 0.99, and 0.99, respectively. For the CT image dataset, these metrics are 96.18%, 96.18%, 96.18%, 0.9618, and 0.96, respectively. Our methodology bridges the gap between precision and interpretability in clinical settings by combining model diversity with explainability, promising enhanced disease diagnosis and greater clinician acceptance.