Dynamic graph representation learning has become essential for analyzing evolving networks in domains such as social network analysis, recommendation systems, and traffic analysis. However, existing continuous-time methods face three key challenges: (1) some methods depend solely on node-specific memory without effectively incorporating information from neighboring nodes, resulting in embedding staleness; (2) most fail to explicitly capture correlations between node neighborhoods, limiting contextual awareness; and (3) many fail to fully capture the structural dynamics of evolving graphs, especially in absence of rich link attributes. To address these limitations, we introduce TAWRMAC—a novel framework that integrates Temporal Anonymous Walks with Restart, Memory Augmentation, and Neighbor Co-occurrence embedding. TAWRMAC enhances embedding stability through a memory-augmented GNN with fixed-time encoding and improves contextual representation by explicitly capturing neighbor correlations. Additionally, its Temporal Anonymous Walks with Restart mechanism distinguishes between nodes exhibiting repetitive interactions and those forming new connections beyond their immediate neighborhood. This approach captures structural dynamics better and supports strong inductive learning. Extensive experiments on multiple benchmark datasets demonstrate that TAWRMAC consistently outperforms state-of-the-art methods in dynamic link prediction and node classification under both transductive and inductive settings across three different negative sampling strategies. By providing stable, generalizable, and context-aware embeddings, TAWRMAC advances the state of the art in continuous-time dynamic graph learning.

Deep learning techniques have made it much easier to generate realistic fake content by superimposing or replacing existing images, videos, or audio with highly realistic alternative content. These manipulations often involve the faces or voices of individuals, creating convincing but fabricated representations. Due to the potential misuse of deepfakes for malicious purposes including spreading misinformation, creating fraudulent content, stealing people’s identity, and manipulating public opinion, the development of detection techniques and policies to mitigate harmful effects of deepfakes has become an important research area. In this paper, we combine both spatial and wavelet features to develop a simple yet effective model to detect deepfakes. Specifically, we pass the input color image through the first convolutional layer and employ a one-level wavelet transform to decompose the channel-wise sum of each batch of features. We then thresh-hold the subbands and reconstruct a single channel feature map, which is concatenated with the original batch of features and passed onto the subsequent layers to capture the facial manipulations using both spatial and frequency features. The expanded wavelet transformed features are fed into the VGG19 backbone to help detect deepfakes with an improved detection performance. We perform both within and cross domain evaluations to compare the performance of the proposed model and state-of-the-art peer models in terms of Area Under Curve (AUC) and Equal Error Rate (EER) metrics. Our extensive experimental results demonstrate that the proposed wavelet-integrated VGG19 model offers a more robust solution than the peer wavelet-integrated Xception model and both VGG19 and Xception baseline models to combating the proliferation of fake multimedia content on digital platforms.

Deep learning techniques have made it much easier to generate realistic fake content by superimposing or replacing existing images, videos, or audio with highly realistic alternative content. Due to the potential misuse of deepfakes for malicious purposes, the development of detection techniques and policies to mitigate harmful effects of deepfakes has become an important research area. In this paper, we combine both spatial and multi-resolution wavelet features to develop a simple yet effective model to detect deepfakes. We perform cross domain evaluations to compare the performance of the proposed model and state-ofthe-art peer models in terms of Area Under Curve (AUC) and Equal Error Rate (EER) metrics. Our extensive experimental results demonstrate that the proposed multi-resolution VGG19 model offers a more robust solution than other compared models to combating the proliferation of fake multimedia content on digital platforms.

The access to large-scale public databases along with the fast progress of deep learning techniques have led to the generation of very realistic fake content. This has raised significant concerns because of their use in social media and the generation of fake news. Thus, the detection of such manipulations has become an increasingly important research area, especially the detection of fake faces has become very important in the field of digital forensics. This paper presents a complementary attention-based deep learning system to detect fake faces. This system effectively incorporates our proposed simple Layer-Integrated Channel Attention (LICA) and Scaled Spatial Attention (SSA) mechanisms in VGG network architecture to capture the importance along each channel and at each spatial location to distinguish between real and manipulated faces and improve detection performance. Our extensive experimental results demonstrate that the proposed system outperforms the state-of-the-art system in detecting fake faces generated by each of the four commonly used manipulations including entire face synthesis, identity swap, attribute manipulation, and expression swap in terms of both accuracy and Area Under Curve (AUC) metrics. It also achieves better performance than the state-of-the-art system to detect fake faces generated by any of the four aforementioned manipulations.

In recent years, Massive Open Online Courses (MOOCs) have gained significant traction as a rapidly growing phenomenon in online learning. Unlike traditional classrooms, MOOCs offer a unique opportunity to cater to a diverse audience from different backgrounds and geographical locations. Renowned universities and MOOC-specific providers, such as Coursera, offer MOOC courses on various subjects. Automated assessment tasks like grade and early dropout predictions are necessary due to the high enrollment and limited direct interaction between teachers and learners. However, current automated assessment approaches overlook the structural links between different entities involved in the downstream tasks, such as the students and courses. Our hypothesis suggests that these structural relationships, manifested through an interaction graph, contain valuable information that can enhance the performance of the task at hand. To validate this, we construct a unique knowledge graph for a large MOOC dataset, which will be publicly available to the research community. Furthermore, we utilize graph embedding techniques to extract latent structural information encoded in the interactions between entities in the dataset. These techniques do not require ground truth labels and can be utilized for various tasks. Finally, by combining entity-specific features, behavioral features, and extracted structural features, we enhance the performance of predictive machine learning models in student assignment grade prediction. Our experiments demonstrate that structural features can significantly improve the predictive performance of downstream assessment tasks. The code and data are available in https://github.com/DSAatUSU/MOOPer_grade_prediction

We propose to combine short-term block-based fuzzy support vector machine (FSVM) learning and long-term dynamic semantic clustering (DSC) learning to bridge the semantic gap in content-based image retrieval. The short-term learning addresses the small sample problem by incorporating additional image blocks to enlarge the training set. Specifically, it applies the nearest neighbor mechanism to choose additional similar blocks. A fuzzy metric is computed to measure the fidelity of the actual class information of the additional blocks. The FSVM is finally applied on the enlarged training set to learn a more accurate decision boundary for classifying images. The long-term learning addresses the large storage problem by building dynamic semantic clusters to remember the semantics learned during all query sessions. Specifically, it applies a cluster-image weighting algorithm to find the images most semantically related to the query. It then applies a DSC technique to adaptively learn and update the semantic categories. Our extensive experimental results demonstrate that the proposed short-term, long-term, and collaborative learning methods outperform their peer methods when the erroneous feedback resulting from the inherent subjectivity of judging relevance, user laziness, or maliciousness is involved. The collaborative learning system achieves better retrieval precision and requires significantly less storage space than its peers. © 2011 Wiley Periodicals, Inc.

Summary form only given. A differentiation-based multi-resolution approach for lossless image compression was presented. The technique used six appropriately weighted adjacent pixels to estimate the predicted pixel intensity. It can also be considered as an additional rule for the seven predictors used by lossless JPEG. The prediction error was decomposed by integer wavelet transform to improve the prediction. The minimum weighted entropy determines the optimal decomposition level of the differentiation-based wavelet transform. The performance of the proposed approach was evaluated and compared with the lossless JPEG and lossless adaptive linear predictor schemes. These two techniques are capable of estimating the present pixel value from the previous pixel values. The computational complexities were greatly reduced due to known values of the predictor coefficients used by the encoder and decoder.