UNIST Develops AI Technology to Enhance Accuracy in Semiconductor Process Predictions

A groundbreaking advancement in semiconductor technology has emerged from the Ulsan National Institute of Science and Technology (UNIST), with a research team led by Professor Jeong Chang-wook developing a novel AI-driven algorithm. This innovation promises to significantly improve the precision of predicting heat distribution and stress points within the intricate processes of semiconductor manufacturing and packaging.

The newly developed algorithm acts as a sophisticated data correction tool, offering enhanced accuracy in forecasting how heat will spread and where mechanical forces will concentrate. Such precise predictions are critical for optimizing production, preventing defects, and ultimately improving the reliability and performance of semiconductor chips, a cornerstone of modern technology.

The significance of this achievement is underscored by its acceptance into the International Conference on Learning Representations (ICLR) 2026, recognized as one of the top three international conferences in the field of artificial intelligence. This international recognition highlights the global impact and potential of UNIST's research, positioning South Korea at the forefront of AI innovation in the crucial semiconductor industry. From a Korean perspective, this development is not just a scientific breakthrough; it represents a strategic leap forward in maintaining our competitive edge in a sector vital to our national economy.

“

A new artificial intelligence (AI)-based data correction algorithm that can predict the path of heat diffusion and the location of concentrated force during semiconductor processes or packaging.