Knowledge Augmented

Knowledge augmentation enhances machine learning models by integrating external knowledge sources or activating internal knowledge within models like large language models (LLMs). Current research focuses on methods to effectively incorporate this knowledge, including techniques like generating synthetic data, refining existing knowledge bases (e.g., medical reports), and developing novel prompt tuning frameworks that leverage both internal and external knowledge. This approach addresses limitations of purely data-driven models, improving performance, reliability, and sample efficiency across various applications, such as question answering, medical image analysis, and autonomous driving, by mitigating issues like hallucinations and improving generalization.