The Ateneo Laboratory for Intelligent Visual Environments (ALIVE) and international researchers have developed a deep learning model that aims to revolutionize dentistry, with the capability to identify tooth and sinus structures in dental X-rays with an accuracy of 98.2%.

This study introduces the stereoisomerism to inform the balanced selection of electrolyte additives to achieve the optimal deposition behaviors and electrochemical performance. An appropriate level of solvation ability and adsorption ability can facilitate faster deposition kinetics and encourage the preferential orientation growth of (002) planes. With sorbitol, the symmetrical battery can stably cycle for over 3470 h at 2 mA cm^-2 and 1 mAh cm^-2, and full battery also exhibits excellent electrochemical performance.

Computer scientists developed E2URec, an efficient unlearning method for LLM-based recommenders. It uses a lightweight module and dual teachers to forget specific data while maintaining performance, which innovates privacy handling in recommendation systems.

Researchers from California State University Northridge (CSUN), National University of Singapore (NUS), NASA Jet Propulsion Laboratory (JPL), and University of Wisconsin-Madison (UW-Madison) have introduced a new concept called autonomous additive manufacturing (AAM), where AI agents take over tasks traditionally managed by human operators. This breakthrough represents a major step toward creating autonomous manufacturing systems, offering improvements in knowledge representation and multi-modal capabilities in additive manufacturing (AM) processes.

The lead Ph.D. candidate, Mr. Haolin Fan, explained: "In the era of generative AI, this research points out a future where human expertise and AI collaborate seamlessly, leading to more resilient and adaptable manufacturing systems that could transform industrial production."

Research team proposed RSLR, a robust self-training approach with label refinement for unsupervised domain adaptation under label noise. It uses LNet for pseudo-labeling and TNet for target-specific training, achieving successful performance on benchmark datasets.