Exploring the intersection of neural networks and symbolic reasoning
Bridging neural networks and symbolic reasoning to build hybrid, interpretable systems that combine structured knowledge with deep learning.
Developing reliable citation techniques and evidence selection methods that ensure every statement in generated text is traceable to its origin.
Building AI systems that are transparent, robust and fair, focusing on interpretability, safety and user confidence in automated decisions.
Designing efficient retrieval and ranking algorithms to fetch relevant documents and passages that power retrieval-augmented generation.
Exploring natural language processing from sentiment analysis to language generation, with a focus on large language models and reasoning.