Title : Brain-inspired Electronic Devices for Low-power Artificial Intelligence
일시 : 2026년 4월 7일 (화), 17시
장소 : 아산이학관 433호
Speaker : 왕건욱 (고려대 융합대학원)
Abstracts:
For sustainable advancements in electronics technology, the field of neuromorphic electronics, i.e., electronics that imitate the principle behind biological synapses with a high degree of parallelism, has recently emerged as a promising candidate for novel computing technologies. In this presentation, I will introduce a three-terminal vertical organic ferroelectric barristor equipped with synaptic functions based on Schottky barrier height modulation to implement a neural network with parallel concurrent execution for static information. The array enables fast and energy-efficient operation of a diagonal convolutional neural network (CNN) that performs simultaneous weight update of cells sharing a kernel matrix, demonstrating the reduction in the energy consumption by ~80 %, compared with the use of a conventional CNN structure. Then, I will present a wide reservoir computing system that is an advanced architecture composed of multiple reservoir layers in parallel, which enables more complex and diverse internal dynamics for multiple time-series information processing. Our approach can efficiently capture and process 3D local dynamic features of multiple time series by implementing a 3D stacked 3×10×10 tungsten oxide (WOx) array architecture. We demonstrate reliable reservoir operations under various time-dependent electrical inputs without overlapping reservoir states. As a proof of concept, we perform spatiotemporal pattern classification of biological cell positions and behavior prediction of a time-dependent chaotic Lorenz attractor. Our proposal achieves a higher and faster (~4.5 times) classification accuracy for 3D positions of a biological cell using 60% fewer memristor cells, and a tenfold reduction in prediction error for the chaotic Lorenz attractor compared with the use of a single reservoir. If time permits, I will also present the recent approaches in materials and devices for neuromorphic electronic applications.