Temporal link prediction (TLP) is a prominent problem in network analysis that focuses on predicting the existence of future connections or relationships between entities in a dynamic network over time. The predictive capabilities of existing models of TLP are often constrained due to their difficulty in adapting to the changes in dynamic network structures over time. In this article, an improved TLP model, denoted as TLP-NEGCN, is introduced by leveraging network embedding, graph convolutional networks (GCNs), and bidirectional long short-term memory (BiLSTM). This integration provides a robust model of TLP that leverages historical network structures and captures temporal dynamics leading to improved performances. We employ graph embedding with self-clustering (GEMSEC) to create lower dimensional vector representations for all nodes of the network at the initial timestamps. The node embeddings are fed into an iterative training process using GCNs across timestamps in the dataset. This process enhances the node embeddings by capturing the network’s temporal dynamics and integrating neighborhood information. We obtain edge embeddings by concatenating the node embeddings of the end nodes of each edge, encapsulating the information about the relationships between nodes in the network. Subsequently, these edge embeddings are processed through a BiLSTM architecture to forecast upcoming links in the network. The performance of the proposed model is compared against several baselines and contemporary TLP models on various real-life temporal datasets. The obtained results based on various evaluation metrics demonstrate the superiority of the proposed work.

In recent years, the study of artificial intelligence (AI) has undergone a paradigm shift. This has been propelled by the groundbreaking capabilities of generative models both in supervised and unsupervised learning scenarios. Generative AI has shown state-of-the-art performance in solving perplexing real-world conundrums in fields such as image translation, medical diagnostics, textual imagery fusion, natural language processing, and beyond. This paper documents the systematic review and analysis of recent advancements and techniques in Generative AI with a detailed discussion of their applications including application-specific models. Indeed, the major impact that generative AI has made to date, has been in language generation with the development of large language models, in the field of image translation and several other interdisciplinary applications of generative AI. Moreover, the primary contribution of this paper lies in its coherent synthesis of the latest advancements in these areas, seamlessly weaving together contemporary breakthroughs in the field. Particularly, how it shares an exploration of the future trajectory for generative AI. In conclusion, the paper ends with a discussion of Responsible AI principles, and the necessary ethical considerations for the sustainability and growth of these generative models.

Exposure to half-truths or lies has the potential to undermine democracies, polarize public opinion, and pro-mote violent extremism. Identifying the veracity of fake news is a challenging task in distributed and disparate cyber-socio platforms. To enhance the trustworthiness of news on these platforms, in this article, we put forward a fake news detection model, OptNet-Fake. The proposed model is architecturally a hybrid that uses a meta-heuristic algorithm to select features based on usefulness and trains a deep neural network to detect fake news in social media. The d-D feature vectors for the textual data are initially extracted using the term frequency inverse document frequency (TF-IDF) weighting technique. The extracted features are then directed to a modified grasshopper optimization (MGO) algorithm, which selects the most salient features in the text. The selected features are then fed to various convolutional neural networks (CNNs) with different filter sizes to process them and obtain the n-gram features from the text. These extracted features are finally concatenated for the detection of fake news. The results are evaluated for four real-world fake news datasets using standard evaluation metrics. A comparison with different meta-heuristic algorithms and recent fake news detection methods is also done. The results distinctly endorse the superior performance of the proposed OptNet-Fake model over contemporary models across various datasets.

Link prediction (LP) plays a crucial role in network science, which forecasts potential connections or relationships between nodes or entities within the network. Link prediction has found many applications, such as suggesting new connections to users in social networks, personalized recommendations in e-commerce, predicting new routes in transportation networks, and many others. This paper introduces a deep learning-based link prediction model in social networks by leveraging graphSAGE (graph sample and aggregation) and Variational Autoencoders (VAE). The proposed work starts by utilizing a graphSAGE to generate node embeddings of the input network by sampling and aggregating information from neighborhood nodes. The generated embeddings are sufficiently expressive to capture the local and global network structure. Further, we adopt VAE to learn a latent space representation of the graphSAGE embeddings. The VAE helps to refine the node embeddings and learn a meaningful latent space representation of the input data, which can be useful for the downstream link prediction task. The encoder’s output (latent space) can capture important features that aid link prediction. Finally, we train a logistic regression classifier using the latent representations from the VAE as features to predict the upcoming links in the network. The necessary hyperparameter studies are performed to obtain the optimal values of the various model parameters. The experiments and simulations conducted on eight different real-world network datasets illustrate the effectiveness of the proposed link prediction model. Additionally, we evaluate the average performance of each comparative link prediction method across all datasets to assess the efficacy of the proposed approach.

Online social media allows users to connect with a large number of people across the globe and facilitate the exchange of information efficiently. These platforms cater to many of our day-to-day needs. However, at the same time, social media have been increasingly used to transmit negative stances such as derogatory language, hate speech, and cyberbullying. The task of identifying the negative stances from social media posts or comments or tweets is termed negative stance detection . One of the major challenges associated with negative stance detection is that most of the content published on social media is often in a multilingual format. This work aims to identify negative stances from multilingual data streams in low-resource languages on social media using a hybrid transfer learning and deep convolutional neural network approach. The proposed work starts by preprocessing the multilingual datasets by removing irrelevant information such as special characters and hyperlinks. The processed dataset is then passed through a pretrained BERT (bidirectional encoder representations from Transformers) model to generate embeddings by fine-tuning the model as per the dataset under consideration. The generated word embeddings are then passed to a deep convolutional neural network for extracting the latent features from the texts and removing the unessential information. This helps our model to achieve robustness and effectiveness for efficient learning on the given dataset and make appropriate predictions on zero-shot data. The article utilizes several optimization strategies for examining the impact of fine-tuning different BERT layers on the model’s performance. Intensive experiments on a variety of languages — namely, English, French, Italian, Danish, Arabic, Spanish, Indonesian, German, and Portuguese — are performed. The experimental results demonstrate the effectiveness and efficiency of the proposed framework.

The rise of location-based services has led to the widespread adoption of location-based social networks (LBSNs), which play a vital role in making recommendations for the next Point-of-Interest (POI). This paper introduces a modified node2Vec and attention-based fusion framework for the next POI recommendation. We start by preprocessing the raw data to gather the relevant information and present a modified node2vec algorithm to generate the feature vectors for users and locations. These feature vectors are then processed using the attention-based framework. The processed features are then used to create well-labeled and balanced datasets which are grouped by specific time intervals. These datasets are then used for training various ML classifiers which are ensembled in a weighted manner to make an improved fusion based recommendation system. The intensive experimental simulations demonstrate the effectiveness of the proposed framework over existing state-of-art methods.

The integration of Natural Language Processing (NLP) and machine learning technologies into the field of mental health and stress prediction represents a significant advancement in the early detection and intervention of mental health issues. This paper explores the application of various NLP and transformer-based techniques, such as TF-IDF, n-grams, BERT, RoBERTa, and ALBERT, to analyze textual data from diverse sources such as social media platforms towards mental health and stress prediction. The developed methodology involves comprehensive data preprocessing, innovative feature extraction, and the application of deep learning models to understand the refinement of language indicative of mental health states. By comparing the effectiveness of different models and approaches, this paper aims to establish a robust framework for accurately predicting mental health conditions, thereby contributing to the broader goal of improving mental health care through technology. The simulation performed on a real-life dataset obtained from Reddit reveals that RoBERTa-based techniques perform better than the other methods.

Nowadays, users across the globe interact with one another for information exchange, communication, and association on various online social media. However, some individuals exploit these venues for malicious practices like hate speech and cyberbully. In this paper, we present an improved multilingual hate speech and cyberbully detection model using bagging-stacking based hybrid ensemble deep learning techniques. The proposed model utilizes Bi-directional Long Short-Term Memory (BiLSTM), Bi-directional Gated Recurrent Unit (Bi-GRU), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM) techniques to enhance the overall performance. We first preprocess the multilingual data streams followed by adoption of Global vectors for word Representation (GloVe) embeddings to convert words to a vector representation in parallel enabling the data streams for binary classification task. In order to construct an architecture for the detection of hate speech and cyberbully, we introduce a heterogeneous fusion of multiple effective models in a unique approach such that CNN-LSTM utilizes a stacking approach with stochastic gradient descent to achieve optimal weights, whereas all the base learners used bagging ensemble approach with cross-validation to reach optimal weights. The final output layer of the proposed ensemble deep learning architecture is achieved using a super learner approach on base learners. To show the efficacy of the proposed model, we conduct the simulation on a total of nine real-world social media datasets in different languages and compared the results with other contemporary hate speech and cyberbully detection methods. The collected findings show that the proposed model outperforms other models on considered datasets and shows an improvement of at least 4.44% in F1 scores.

This paper introduces an approach for Activity Recognition by integrating channel-wise attention with Motion of Oriented Gradients (MOG) and Appearance Information (AI) alongside a point-wise ConvBi-LSTM. The proposed model features a dual-stream architecture to process RGB and Gradient Motion Information (GMI) frames, extracting appearance information and motion dynamics using the Xception network. The first stream focuses on RGB frames to capture appearance details, while the second stream uses GMI frames to extract motion dynamics. The Xception network is employed to derive appearance and motion features from RGB and GMI frames combined to generate comprehensive insights. The model further integrates channel-wise attention and Bi-LSTM through point-wise convolutional connections to identify and prioritise highly informative features for accurate classification. Extensive experiments on benchmark datasets, UCF Crime, HBD21, XD Violence and a synthesized dataset demonstrated the model's effectiveness, achieving accuracies of 93.71%, 95.1%, 88.8%, and 96.4%, respectively.

Humour is a crucial aspect of human speech, and it is, therefore, imperative to create a system that can offer such detection. While data regarding humour in English speech is plentiful, the same cannot be said for a low-resource language like Hindi. Through this paper, we introduce two multimodal datasets for humour detection in the Hindi web series. The dataset was collected from over 500 minutes of conversations amongst the characters of the Hindi web series Kota-Factory and Panchayat . Each dialogue is manually annotated as Humour or Non-Humour. Along with presenting a new Hindi language-based Humour detection dataset, we propose an improved framework for detecting humour in Hindi conversations. We start by preprocessing both datasets to obtain uniformity across the dialogues and datasets. The processed dialogues are then passed through the Skip-gram model for generating Hindi word embedding. The generated Hindi word embedding is then passed onto three convolutional neural network (CNN) architectures simultaneously, each having a different filter size for feature extraction. The extracted features are then passed through stacked Long Short-Term Memory (LSTM) layers for further processing and finally classifying the dialogues as Humour or Non-Humour. We conduct intensive experiments on both proposed Hindi datasets and evaluate several standard performance metrics. The performance of our proposed framework was also compared with several baselines and contemporary algorithms for Humour detection. The results demonstrate the effectiveness of our dataset to be used as a standard dataset for Humour detection in the Hindi web series. The proposed model yields an accuracy of 91.79 and 87.32 while an F1 score of 91.64 and 87.04 in percentage for the Kota-Factory and Panchayat datasets, respectively.

Finding influential nodes reshuffles the very notion of linear paths in business processes and replaces it with networks of business value within a smart enterprise system. There are many existing algorithms for identifying influential nodes with certain limitations for applying in large-scale networks. In this article, we propose a community structure with integrated features ranking (CIFR) algorithm to find influential nodes in the network. First, we use the community detection algorithm to find communities in the system, and then we rank the nodes of network based on three factors, namely local ranking, gateway ranking, and community ranking, collectively termed as integrated features. Our algorithm intends to select influential nodes, which are both globally and locally optimal, leading to overall high information propagation. We perform the experimental results on total eight networks using various evaluation parameters. The obtained results validate superior performance against contemporary algorithms adding value to smart enterprises.

Over the last couple of decades, Social Networks have connected people on the web from across the globe and have become a crucial part of our daily life. These networks have also rapidly grown as platforms for propagating products, ideas, and opinions to target a wider audience. This calls for the need to find influential nodes in a network for a variety of reasons, including the curb of misinformation being spread across the networks, advertising products efficiently, finding prominent protein structures in biological networks, etc. In this paper, we propose Modified Community Diversity (MCD), a novel method for finding influential nodes in a network by exploiting community detection and a modified community diversity approach. We extend the concept of community diversity to a two-hop scenario. This helps us evaluate a node's possible influence over a network more accurately and also avoids the selection of seed nodes with an overlapping scope of influence. Experimental results verify that MCD outperforms various other state-of-the-art approaches on eight datasets cumulatively across three performance metrics.

The Asian social networks are dominated by the society’s collectivist culture, and this interestingly introduces an influence mechanism aided by word-of-mouth and opinion leaders. An opinion leader can help to generate and shape other people’s opinion and achieve a high information spread on any topic. In this work, a modified spider monkey optimization based opinion leader detection approach is proposed. Firstly, we employ the modified node2vec graph embedding to generate the lower dimensional vectors which act as the initial features for the nodes in a typical Asian social network. Next, the entire population is broken down into several groups using the k-means++ algorithm where the number of clusters is equal to the number of opinion leaders to be selected. The local and global leaders are chosen by using the coordinates of the cluster centres of these clusters. The coordinates of the centroids of the clusters are then used to detect the local and global leaders in the network. The local leaders then form the seed set of opinion leaders for the network. The positions of the nodes in the network, including the local and global leaders, are updated over a number of iterations. At the end of these iterations, the seed set generating the maximum influence forms the set of opinion leaders in the network. We test our proposed approach using the popular information diffusion and cognitive opinion dynamics (COD) models. We perform intensive experiments on several real-life social networks based on various performance metrics. The results obtained reveal that the proposed approach outperforms several existing techniques of opinion leader detection.

The lack of sufficient contact opportunities in sparse vehicular delay tolerant networks results in poor performance. The network performance can be enhanced by deploying additional small wireless devices called throwbox. Nonetheless, the deployment of throwboxes with distinct placement strategies results in diverse outcomes in terms of metrics such as delivery ratio (DR), delay, and overhead cost (OC). In this paper, the throwbox deployment problem has been presented as an optimization problem. A customized binary differential evolution (CBDE)-based iterative optimization method is employed to improve the DR. Furthermore, depending upon a specific application, it might be desired to have a balanced performance that optimizes the trade-off between different metrics. Therefore, the employment of a multi-objective CBDE (MOCBDE) technique is proposed to optimize the trade-off between DR and OC metrics. The proposed approaches are implemented using Matlab and opportunistic network environment simulator. The experimental results presented in the paper validates the superiority of the proposed CBDE approach over existing methods of throwbox deployment. In addition, the MOCBDE approach provides a set of non-dominated (also called pareto-optimal) solutions with an exhaustive view of desirable results. Consequently, depending on the application requirements, the method allows selecting suitable locations for throwbox deployment.

Electrocardiogram (ECG) is a widely used technique to diagnose cardiovascular diseases. It is a non-invasive technique that represents the cyclic contraction and relaxation of heart muscles. ECG can be used to detect abnormal heart motions, heart attacks, heart diseases, or enlarged hearts by measuring the heart’s electrical activity. Over the past few years, various works have been done in the field of studying and analyzing the ECG signals to detect heart diseases. In this work, we propose a deep learning and fuzzy clustering (Fuzz-ClustNet) based approach for Arrhythmia detection from ECG signals. We started by denoising the collected ECG signals to remove errors like baseline drift, power line interference, motion noise, etc. The denoised ECG signals are then segmented to have an increased focus on the ECG signals. We then perform data augmentation on the segmented images to counter the effects of the class imbalance. The augmented images are then passed through a CNN feature extractor. The extracted features are then passed to a fuzzy clustering algorithm to classify the ECG signals for their respective cardio diseases. We ran intensive simulations on two benchmarked datasets and evaluated various performance metrics. The performance of our proposed algorithm was compared with several recently proposed algorithms for heart disease detection from ECG signals. The obtained results demonstrate the efficacy of our proposed approach as compared to other contemporary algorithms.

The recent Coronavirus disease (COVID-19), which started in 2019, has spread across the globe and become a global pandemic. The efficient and effective COVID-19 detection using chest X-rays helps in early detection and curtailing the spread of the disease. In this paper, we propose a novel Trained Output-based Transfer Learning (TOTL) approach for COVID-19 detection from chest X-rays. We start by preprocessing the Chest X-rays of the patients with techniques like denoising, contrasting, segmentation. These processed images are then fed to several pre-trained transfer learning models like InceptionV3, InceptionResNetV2, Xception, MobileNet, ResNet50, ResNet50V2, VGG16, and VGG19. We fine-tune these models on the processed chest X-rays. Then we further train the outputs of these models using a deep neural network architecture to achieve enhanced performance and aggregate the capabilities of each of them. The proposed model has been tested on four recent COVID-19 chest X-rays datasets by computing several popular evaluation metrics. The performance of our model has also been compared with various deep transfer learning models and several contemporary COVID-19 detection methods. The obtained results demonstrate the efficiency and efficacy of our proposed model.

Nowadays, social media and virtual networking hubs like Twitter, Facebook have become an integral part of our daily lives. The recent boom in multimedia technology and increased internet access has led us into a hyper-connected global world. But these networks are often observed as the conduit for propagating fake news, which may cause a severe problem to a healthy social environment and destroy the harmony between the users. This calls for a proper segregation tool to classify various news articles as real or fake. Numerous research has been done on this topic, including the use of Artificial Intelligence (AI). In this work, we propose a deep learning based hybrid framework utilizing Word2Vec embedding and LSTM for fake news detection. As part of our approach, we generate Word2Vec embedding for obtaining vector representations of the news excerpts. The Word2Vec embeddings assist in generating context-free and data agnostic feature vectors for our news articles. The stacked LSTM layers process the extracted feature vectors to obtain the topic-relevant salient features for the news articles. This is followed by two fully connected dense layers for classifying whether the news excerpt under consideration is real or fake. We also perform hyperparameter tuning for achieving a better performance of our model. The proposed model is context-free and independent of datasets as well as topics for fake news detection. We compare the proposed method’s performance with some traditional Machine Learning baseline models, deep learning models, the pre-trained Bidirectional Encoder Representations from Transformers (BERT) via transfer learning, and some recently proposed state-of-the-art models. These models are tested on four datasets belonging to different domains for both training and testing purposes. Our proposed technique outperforms other well-known methods based on various performance metrics through intensive experimentation.

Abnormal activity detection is an area of study that focuses on finding abnormal activities like theft, vandalism, abuse, etc in video and CCTV footage. In this paper, a combination of skeleton based and transformer-based approach is used. The skeletons are used to enrich the motion of human bodies in the video. The transformer is used to model dynamic convolutions which helps to overcome the problems associated with fixed size kernels. To preserve the background information the architecture also uses feature extraction. The methodology used in the paper, also proposed a way to use transfer learning in the transformer network. The proposed framework provides excellent results outperforming many of existing methodologies. ROC (AUC score) attained by the proposed method for some classes of actions are up to 97.71%, which is even higher than current state of art approaches. The average ROC (AUC score) of our method is better than many of the approaches which are considered in the paper.

Social networks have emerged as efficient platforms to connect people worldwide and facilitate the rapid spread of information. Identifying influential nodes in social networks to accelerate the spread of particular information across the network formulates the study of influence maximization (IM). In this paper, inspired by deep learning techniques, we propose a novel approach to solve the influence maximization problem as a classical regression task using transfer learning via graph-based long short-term memory (GLSTM). We start by calculating three popular node centrality methods as feature vectors for the nodes in the network and every node's individual influence under susceptible-infected-recovered (SIR) information diffusion model, which forms the labels of the nodes in the network. The generated feature vectors and their corresponding labels for all the nodes are then fed into a graph-based long short-term memory (GLSTM). The proposed architecture is trained on a vast and complex network to generalize the model parameters better. The trained model is then used to predict the probable influence of every node in the target network. The proposed model is compared with some of the well-known and recently proposed algorithms of influence maximization on several real-life networks using the popular SIR model of information diffusion. The intensive experiments suggest that the proposed model outperforms these well-known and recently proposed influence maximization algorithms.

Online social networks have become a consistent part of our day-to-day life. They virtually connect people around the world and serve as ideal platforms for interactions, sharing of information, ideas, and products. Influence maximization (IM) is the problem of maximizing the reach of an idea or an opinion in a network by shortlisting the most influential nodes in the respective network, which are further used as seed nodes to spread the information in the rest of the network. It is a problem of great relevance in today’s world because of its real life applicability in the business. Numerous methods have been proposed in the literature to rank the nodes according to their spreading ability and certain other characteristics. In this paper, we propose a novel method to solve the problem of influence maximization named Communities based Spreader Ranking (CSR), which is based on the notions of communities and bridge nodes. It identifies bridge nodes as influential nodes based on three concepts: community diversity, community modularity, and community density. Community diversity is used to identify bridge nodes and the rest two are used to identify significant communities. Extensive experimentation validation on various datasets using popular information diffusion models demonstrates that the proposed method delivers proficient results compared to numerous previously known contemporary influence maximization methods.