E-MIA: Black-Box Membership Inference Attacks on RAG Systems
Explore E-MIA, a novel exam-style black-box membership inference attack improving data privacy risks in Retrieval-Augmented Generation (RAG) systems.
Graph Rewiring in GNNs to Fix Over-Squashing & Smoothing
Explore graph rewiring techniques to overcome over-squashing and over-smoothing in GNNs, enhancing performance and information flow in deep networks.
CodeFP: Advanced Co-Generative De Novo Protein Design
Discover how CodeFP revolutionizes de novo protein design by improving functional consistency and foldability for biotech and medical applications.
SCARV: Stable Sample Ranking for Redundant NLP Data
Discover SCARV, a framework enhancing stable sample ranking in redundant NLP datasets by tackling duplicates and paraphrases for reliable data analysis.
Interpretable Experiential Learning for Smarter AI Models
Discover a novel interpretable experiential learning model that enhances reinforcement learning with state history and global feedback for better AI decisi...