Relational Learning

Relational learning focuses on modeling and leveraging relationships between objects within data, aiming to improve machine learning performance on tasks where these relationships are crucial. Current research emphasizes developing novel architectures, such as graph neural networks and variations of transformer models incorporating attention mechanisms, to effectively capture complex relational structures in diverse data types, including tabular data, knowledge graphs, and multimodal information. This field is significant because it enables more accurate and efficient predictions in various applications, ranging from drug discovery and battery life prediction to knowledge graph completion and text-to-image generation, by explicitly incorporating relational context.