Statistical Relational

Statistical Relational Learning (SRL) combines logic and probability to model and reason with relational data, aiming to improve the representation and inference capabilities of AI systems. Current research focuses on developing efficient inference algorithms, particularly for lifted inference which scales to large datasets, and on improving model architectures like probabilistic relational models and neural-symbolic systems to handle complex relationships and incorporate prior knowledge. These advancements are significant for various applications, including causal inference, knowledge graph completion, and privacy-preserving data synthesis, by enabling more accurate and scalable reasoning over structured data.