Chaeyun Kim

Humanoid robots struggle to bridge the gap between selecting the right tool and executing complex tasks, with current policies showing significant performance degradation on unseen tools and unrelated instructions.

Even leading MLLMs struggle with object hallucination, revealing critical integration-stage weaknesses that standard benchmarks fail to capture.

Financial LLMs can now be rigorously evaluated against targeted risks, reducing critical false negatives in safety assessments from 28 to 12.

A unified action embedding space across diverse humanoid robots boosts performance and interpretability, transforming how we approach robot policy learning.

Mapping LLM attack strategies onto a multiplex network reveals interpretable vulnerability clusters and dramatically improves red teaming efficiency.

RIS models struggle with motion-based queries, but a new data augmentation and contrastive learning approach closes the gap without sacrificing performance on appearance-based descriptions.