Research team from Nanjing University proposed FOCUS, a causal model-based offline RL algorithm, which uses causal structure to improve policy generalization, outperforming baselines in offline settings.

Researchers from Shanghai Jiao Tong University and Huawei Noah’s Ark Lab introduced Laser, a novel framework integrating LLMs into recommender systems (RSs) to enhance sample efficiency. The study shows Laser matches/surpasses traditional models using fewer training samples, addressing data sparsity and latency challenges.

Research team published a comprehensive review on circRNA-disease association prediction, summarizing computational models into four categories and analyzing their strengths and limitations. Challenges like data imbalance and insufficient fusion of biological information are highlighted.

Research team introduced clustered reinforcement learning (CRL), a novel RL framework for efficient exploration in large state spaces or sparse reward environments. CRL combines state novelty and quality through clustering, significantly improving exploration performance in continuous control tasks and Atari games.

Achieved the first spectroscopic observation of hydrogen (H₂) and deuterium (D₂) molecules physically adsorbed within an atomic-scale space known as a picocavity.Employed picometric rotational/vibrational spectroscopy to elucidate their structure and dynamics at the single-molecule level.Observed distinct spectral responses for H₂ and D₂, and theoretically demonstrated that these differences arise from non-trivial isotope effects due to quantum nuclear effects.

This advancement in precision molecular spectroscopy within picocavities opens new possibilities for well-controlled studies of functional materials for energy applications, such as hydrogen storage systems and catalytic surfaces, as well as for developing single-molecule quantum control technologies.

The research team—Dr. Hyun Kim at the Korea Research Institute of Chemical Technology (KRICT), Prof. Habeom Lee at Pusan National University, and Prof. Taylor H. Ware at Texas A&M University—successfully developed artificial muscles based on azobenzene-functionalized semicrystalline liquid crystal elastomers (AC-LCEs) that actuate in response to light.