[Sci-Tech NOW] UNIST slashes robust AI training cost by 80% with quantum computing

From left, Professor Yoon Sung-hwan of Ulsan National Institute of Science and Technology (UNIST) and researcher Lee Hyun-gyu. Provided by UNIST

■ Ulsan National Institute of Science and Technology (UNIST) introduced on the sixth {that a} staff led by Professor Yoon Sung-hwan on the Graduate School of Artificial Intelligence and Professor Kim Joong-heon at Korea University’s School of Electrical Engineering has developed a studying methodology known as “QRIM” that dramatically reduces the computational cost of “robust reinforcement learning” utilizing quantum algorithms. To resolve the computational burden that hinders the appliance of robust reinforcement studying, the analysis staff utilized the quantum computing precept of “superposition.” QRIM can shortly determine the worst-case situation, decreasing the training load of robust AI by as much as 80%. This analysis has been accepted to the International Conference on Machine Learning (ICML) 2026. 

■ Gwangju Institute of Science and Technology (GIST) introduced on the sixth {that a} joint analysis staff led by Professor Yoon Myung-han of the Department of Materials Science and Engineering and Professor Shim Bong-seob of Inha University’s Department of Chemical Engineering has developed eco-friendly digital gadgets that may be decomposed by bugs. To deal with the difficulty of digital waste that’s troublesome to gather after use, the staff mixed montmorillonite (MMT), a clay mineral that bugs can ingest, with PEDOT:PSS, a conductive polymer materials extensively utilized in “organic electrochemical transistors (OECTs),” thereby making a composite materials that provides each degradability and electrical efficiency. The analysis outcomes had been revealed on-line on June 29 within the worldwide journal “ACS Polymer Science & Technology.”