Existing computer systems have separate data processing devices and storage devices, making them inefficient in processing complex data such as AI. A KAIST research team has developed a memristor-based integrated system similar to the way our brain processes information. It is now ready to be applied to a variety of devices, including smart security cameras that can instantly recognize suspicious activity without relying on remote cloud servers, and medical devices that can help analyze health data in real time.
KAIST (President Gwanghyung Lee) announced on the 17th that the joint research team of Professor Choi Choi of the Department of Electrical Engineering and Professor Younggyu Yoon has developed a next-generation neuromorphic semiconductor-based ultra-small computing chip. You can learn and correct your errors on your own.
What’s special about this computing chip is that it can learn and correct errors caused by non-ideal characteristics that were difficult to solve in existing neuromorphic devices. For example, when processing a video stream, the chip learns to automatically separate moving objects from the background and becomes better at this task over time.
This self-learning ability has been demonstrated in real-time image processing by achieving accuracy comparable to ideal computer simulations. The research team’s main achievement is that it went beyond developing brain-like components and completed a system that is both reliable and practical.
The research team developed the world’s first memristor-based integrated system that can adapt to immediate environmental changes and presented an innovative solution that overcomes the limitations of existing technology.
At the center of this innovation is a next-generation semiconductor device called a memristor*. The variable resistance properties of this device can replace the role of synapses in neural networks, allowing data storage and calculations to be performed simultaneously, just like our brain cells.
*Memristor: A compound word of memory and resistor, a next-generation electrical device whose resistance value is determined by the amount and direction of charge flowing between two terminals in the past.
The research team designed a highly reliable memristor that can precisely control resistance changes and developed an efficient system that eliminates complex compensation processes through self-learning. This study is significant in that it experimentally verified the feasibility of commercializing a next-generation neuromorphic semiconductor-based integrated system that supports real-time learning and inference.
This technology will revolutionize the way artificial intelligence is used in everyday devices, allowing AI tasks to be processed locally without relying on remote cloud servers, making them faster, more private, and more energy efficient.
KAIST researchers Jeong Hak-cheon and Han Seung-jae, who led the development of this technology, explained, “This system is like a smart workspace where everything is within reach instead of going back and forth between the desk and filing cabinet.” “This is similar to the way our brains process information, where everything is processed efficiently and at one point at the same time.”
This study was conducted by Hak-cheon Jeong and Seung-jae Han, students in the integrated master’s and doctoral program at KAIST’s Department of Electrical Engineering, as co-first authors, and the research results were published in the online edition of the international journal Nature Electronics on January 1. August 8, 2025.
This research was conducted with support from the National Research Foundation of Korea’s Next-Generation Intelligent Semiconductor Technology Development Project, Outstanding New Researcher Project, PIM AI Semiconductor Core Technology Development Project, and Electronics and Telecommunications Research Institute Research and Development Support Project. Department of Information and Communication Technology Planning and Evaluation.
