Weihai Lu

HiLaR reveals that optimizing user preference modeling through hierarchical latent reasoning can significantly outperform traditional recommendation approaches.

Achieve state-of-the-art multimodal stance detection by having multiple AI agents debate each other, complete with retrieval-augmented context and self-reflection.

By mimicking human reasoning through instance-based analogy, RAMM significantly improves multimodal fake news detection, especially in data-scarce domains.