Mechanically Robust Neuromorphic Computing Enabled by Stretchable Adhesive Ion-Gels
발표자
김윤수 (부산대학교)
연구책임자
심현석 (부산대학교)
초록
내용
Neuromorphic computing leveraging flexible electronic devices has emerged as a pivotal technology for next-generation wearable computing and soft robotics. In this work, we present a stretchable synaptic transistor utilizing a newly engineered ion-gel dielectric, enabling stable PPF performance under 50% tensile strain. In conventional ion-gel based synaptic transistors, the PPF index decreased from 1.5 to 1.1 under 50% tensile strain. In contrast, the proposed synaptic transistor retained a PPF index of 1.38, decreasing only slightly from 1.52 under the same condition. Based on this mechanical stability, MNIST digit classification using an artificial neural network (ANN) yielded an accuracy of 91.8% under 50% tensile strain. In addition, object classification using the CIFAR-10 dataset and convolutional neural network (CNN) achieved an accuracy of 85.6% under 50% tensile strain. The proposed device paves the way for next-generation wearable AI applications.