Recent advancements in graph learning are increasingly focused on enhancing the representation and utility of multimodal and dynamic networks. The integration of multimodal data has shifted the paradigm from traditional text-attributed graphs to more complex multimodal-attributed structures, allowing for improved performance in tasks that require modality interaction. Concurrently, the emergence of topological deep learning is addressing the limitations of conventional graph neural networks by modeling higher-order relationships, which are essential for capturing intricate relational structures. Techniques such as counterfactual data augmentation and role-aware modeling are being employed to bolster the robustness of dynamic network predictions amid evolving structures. Additionally, the exploration of Riemannian geometry is providing a new theoretical foundation for graph representation learning, positioning it as a unifying framework. These developments not only enhance the scalability and adaptability of graph learning methods but also open avenues for solving commercial challenges in diverse fields, from social networks to healthcare analytics.
Recently, the rapid advancement of multimodal domains has driven a data-centric paradigm shift in graph ML, transitioning from text-attributed to multimodal-attributed graphs. This advancement signifi...
In this paper we develop a graph-learning algorithm, MED-MAGMA, to fit multi-axis (Kronecker-sum-structured) models corrupted by multiplicative noise. This type of noise is natural in many application...
The rapid growth and continuous structural evolution of dynamic networks make effective predictions increasingly challenging. To enable prediction models to adapt to complex temporal environments, the...
Topological deep learning has emerged for modeling higher-order relational structures beyond pairwise interactions that standard graph neural networks fail to capture. Although combinatorial complexes...
Real-world dynamic graphs are often directed, with source and destination nodes exhibiting asymmetrical behavioral patterns and temporal dynamics. However, existing dynamic graph architectures largely...
The effectiveness of contrastive learning methods has been widely recognized in the field of graph learning, especially in contexts where graph data often lack labels or are difficult to label. Howeve...
Graphs provide a natural representation of relational structure that arises across diverse domains. Despite this ubiquity, graph structure is typically learned in a modality- and task-isolated manner,...
Text-rich graphs, which integrate complex structural dependencies with abundant textual information, are ubiquitous yet remain challenging for existing learning paradigms. Conventional methods and eve...
Graphs are ubiquitous, and learning on graphs has become a cornerstone in artificial intelligence and data mining communities. Unlike pixel grids in images or sequential structures in language, graphs...