ABSTRACT Modelling the evolution of Alzheimer's disease (AD) requires a thorough spatiotemporal study of longitudinal neuroimaging data. We propose in this paper a novel deep learning framework that uses a parallel combination of Recurrent Neural Networks (RNNs) and Vision Transformers (ViT) to extract temporal disease dynamics and spatial structural changes from serial MRI data. While the RNN evaluates sequential dependencies across timepoints, the ViT branch uses self‐attention to derive hierarchical brain‐region characteristics. A stacked auto‐encoder (SAE) fuses these features into a compact representation, enhancing discriminative power while reducing redundancy. Fully connected layers are given the fused features in order to predict progression and classify AD (CN/MCI/AD). We used the ADNI dataset to test our proposed methodology. In terms of disease stage differentiation, our approach reaches state‐of‐the‐art accuracy of 92.3%. Compared to CNN or RNN‐only models, it considerably improves the prediction of the early conversion of MCI to AD (AUC = 0.94). When processing heterogeneous neuroimaging data, the SAE‐based fusion outperforms attention methods. With potential uses in customised treatment planning, this hybrid approach provides a clinically interpretable tool for longitudinal AD.

Spam messages pose a significant cybersecurity threat, leading to phishing attacks, fraud, and privacy breaches. Traditional spam detection methods, such as rule-based filtering and statistical models, often fail to capture the evolving and complex nature of spam messages. In this paper, we propose an Arabic spam detection model leveraging BERT (Bidirectional Encoder Representations from Transformers), a deep learning-based NLP model. Our approach enhances classification accuracy by utilizing contextual text representations specific to the Arabic language. We preprocess Arabic text using AraBERT tokenization and fine-tune the BERT-based model on a balanced dataset of Arabic spam and ham messages. Experimental results demonstrate that our model achieves high accuracy (98%), outperforming traditional machine learning and deep learning approaches. This research highlights the potential of transformer-based models in Arabic spam filtering, paving the way for more efficient and robust detection systems.

Workplace stress, a widespread issue in modern professional environments, significantly increases the potential for errors and accidents. Timely and precise stress identification is vital for fostering a secure and efficient work environment. This research introduces an innovative, comparative-analysis framework designed for real-time stress detection, utilizing Heart Rate Variability (HRV) as a reliable physiological indicator. Unlike standard heart rate measurements, HRV offers a granular view of the autonomic nervous system (ANS) function, enabling accurate stress evaluation. We implement a comprehensive methodology incorporating a refined preprocessing stage—including the removal of outliers, feature selection, and data normalization—along with a comparative assessment of eight deep recurrent neural network (RNN) architectures. These include vanilla RNN, bidirectional RNN (BiRNN), Gated Recurrent Unit (GRU), bidirectional GRU (BiGRU), standard Long Short-Term Memory network (LSTM), bidirectional LSTM (BiLSTM), Peephole LSTM, and Attention-based LSTM, applied to binary stress classification. Utilizing a dataset of 410,322 HRV records from the SWELL Knowledge Work (SWELL-KW) Dataset, our framework demonstrates exceptional performance, with the BiGRU architecture achieving a test accuracy of 99.51%. This study highlights the effectiveness of advanced temporal modeling and comparative analysis in creating robust stress detection systems for various occupational contexts, thereby enhancing workplace safety.

The pervasive issue of online sexism continues to pose significant challenges, fostering environments characterized by toxicity and perpetuating harmful societal norms. In response, this paper presents an approach for the discovery of sexist statements employing convolutional neural networks (CNNs), Word Embeddings, and data augmentation techniques. Through the fusion of CNNs’ capacity for hierarchical feature extraction with the semantic representations afforded by Word Embeddings, our method achieves exemplary discrimination performance. Additionally, the incorporation of data augmentation enriches the training dataset, thereby augmenting model generalization and resilience. Empirical evaluation on a larger dataset of statements demonstrates the efficacy of our approach, surpassing many baseline approaches in terms of discovery accuracy, precision, recall and F1-score.

Cyberbullying, marked by its persistent and intentional aggression online, yields severe repercussions for its victims, extending beyond immediate distress to long-lasting effects such as heightened anxiety, depression, and social withdrawal. Individuals subjected to Cyberbullying often grapple with diminished self-esteem, compromised academic performance, and strained interpersonal relations. Given the escalating prevalence of this digital menace, there is a pressing need for advanced methodologies to address it effectively. This paper introduces an approach to Cyberbullying detection, integrating techniques such as Bag-of-Words (BoW) and Term Frequency-Inverse Document Frequency (TF-IDF) analyses, along with the parallel processing capabilities of Convolutional Neural Networks (CNNs) and the contextual comprehension provided by Bidirectional Long Short-Term Memory (BiLSTM) networks. Through an experimentation on the latest Ejaz-Choudhury-Razi Cyberbullying dataset, our framework exhibits satisfactory performance in identifying instances of online hostility. These results underscore the potential of our approach to significantly contribute to ongoing efforts aimed at combating Cyberbullying in digital environments.

This article presents a novel approach for accurately classifying COVID-19 medical images using deep learning techniques. the proposed approach utilizes a convolutional neural network (CNN) architecture with two inputs segmented CT scan images obtained through deep Learning U-Net for image segmentation, and X-ray images. the CNN effectively classifies the images into three categories COVID-19, non-coronavirus, and pneumonia. the experimental results of the proposed method using the COVID-QU dataset and CT scans showcase its superior performance compared to other state of the art approaches With an impressive accuracy rate of 98% the method demonstrates its ability to accurately detect COVID-19 cases. additionally, the achieved high accuracy, F1-score, and low loss values further emphasize its effectiveness this framework offers a highly accurate solution for detecting COVID-19 infections in CT scans and X-ray images Thereby enhancing the diagnostic capabilities of healthcare professionals in identifying COVID-19 cases.

Face recognition is one of the most widely used biometrics for identifying people. However, face images suffer from several issues that could affect the achieved results, especially in a crowded environment. Such as facial expression, occlusion, low resolution, noise, illumination and pose variation. In this paper, we propose a robust image representation system for face recognition. First, 3D face data reconstructed from 2D images are used instead of 3D capture. This is accomplished by modeling the difference in the texture map of the 3D aligned input and reference images. Then, fusing shape and texture local binary patterns (LBP) on a mesh for face recognition using the Mesh-LBP. Finally, we used a deep Auto-Encoder to create a compact data representation based on the obtained face images descriptors from the Mesh-LBP. Through experiments conducted on the Multi-PIE and Bosphorus databases, we show that our method is very competitive against state-of-the-art methods.

Abstract Video is considered as one of the most useful and important forms of multimedia data, that is usually used in several applications. Despite its importance, video indexing and retrieval becomes a challenging task. In order to reduce the amount of data and keep only relevant frames, keyframe extraction becomes necessary in a content‐based video retrieval (CBVR) system. In this paper, a keyframe extraction method is proposed based on the face image quality for video surveillance systems. Data is reduced by rejecting frames without faces. Then, face images are clustered by identity. After that, a set of candidate frames is selected to be proceeded. The face quality assessment is based on four metrics including pose estimation, sharpness, brightness and resolution, and the frame with the best face quality is considered as a keyframe. Experimental tests were carried on several datasets in order to prove the efficiency of authors' method compared with state‐of‐the‐art approaches.

Keyframe extraction is one of the basic procedures relating to video retrieval and summary. It consists on presenting an abstract of the video with the most representative frames. This paper presents an efficient keyframe extraction approach based on local description and graph modularity clustering. The first step is to generate a set of candidate keyframes using a windowing rule in order to reduce the data to be examined. After that, detect interest points in these set of images. Then compute repeatability between each two images belonging to the candidate set and stocks these values in a matrix that we called repeatability matrix. Finally, the repeatability matrix is modelled by an oriented graph and we will select keyframes using graph modularity clustering principle. The experiments showed that this method succeeds in extracting keyframes while preserving the salient content of the video. Further, we found good values in term of precision, PSNR and compression rate.

Research concerning the detection of interest points features for local query is particularly rich and many methods have been proposed in the literature. Matching of interest points is an essential step in the interest point detectors evaluation process. There are many existing methods for interest points matching and most of them are related to the detectors parameters. In this paper, we present a new matching method based on the prediction validation principle by matching interest points with a local description and with adding spatial constraints. Harris [4], SIFT [6] and SURF [1] were used to detect interest-point features from satellite images. This paper presents a short evaluation of these detectors on satellite images with different spatial resolutions, describes the experimental setup and the results of the detected interest points used for the comparison of the detection rate and the repeatability. Finally, we determine which detector leads to the best results for local query on satellite images.

This paper proposes a blind and high-capacity watermarking framework for 3D triangle meshes. The basic idea is to use a quantization based approach in the transform domain to embed the secret message. Our proposal can be applied to regular as well as irregular meshes by using irregular wavelet-based analysis. The watermark is inserted in an appropriate resolution level by quantizing the norms of wavelet coefficients vectors. Simulation results show that our watermarking framework is robust to common geometric attacks and can provide relative high data embedding rate whereas keep a relative lower distortion.

Query by visual example (QBVE) has been widely exploited in image retrieval. If starting image is missing, the query by visual thesaurus paradigm allows the user to compose his mental query image through visual patches summarizing the region database. Researches on the human visual system have provided considerable evidence that the color and texture must be processed separately. In this paper, we propose to enrich the paradigm of mental image search by constructing a hierarchical visual thesaurus of the regions provided by a new region labeling criterion into homogeneous and textured regions for boosting the object recognition. The new labeling criterion is based on the spatial dispersion of interest points in the region. Our point based criterion has been validated on a satellite image database. We can prove that our approach is able to retrieve complex concepts better than describing homogeneous and textured regions with the same visual feature.