3 papers indexed
Mehak Dhaliwal, Rasta Tadayon, Andong Hua, Haewon Jeong, Yao Qin
LLMs can perform language-based quantitative prediction from unstructured inputs, but remain susceptible to hallucinations and overconfident errors, making it critical to know not only what a model predicts, but when its predictions can be trusted. We introduce CARE-PPO, a reinfo…
Chapa Sirithunge, Haewon Jeong, Qinghua Guan, Fumiya Iida, Josie Hughes
Nonverbal communication through head and neck movement is fundamental to human social signalling, yet how robotic neck morphology translates motion into communicative information remains poorly understood. We present an information-theoretic framework characterising robot neck mo…
Irina Espejo Morales, Damon Hinz, Marvin Alberts, Geraud Krawezik, Haewon Jeong, Shirley Ho
Structural elucidation from Nuclear Magnetic Resonance (NMR) data remains a fundamental bottleneck across chemistry, materials science, and biology. We demonstrate that an agentic AI system can perform this task at a level comparable to graduate-level chemistry students. Instead…