11320E500100 Seminar

TOPIC
▸　Advancing Neuromorphic Computing for Future Sensory Memory Systems 

❝　Neuromorphic computing, inspired by biological neural systems, is a promising approach for energy-efficient artificial intelligence. This work explores key advancements in neuromorphic systems, focusing on resistive random-access memory (ReRAM), piezotronic devices, and spiking neural networks (SNNs). We first examine the development and application of ReRAM technology, including material studies, process optimization, and its role in analog memory and neuromorphic computing. By utilizing ReRAM as synaptic weights in analog neural networks, we enhance performance through calibration techniques, improving both reliability and efficiency. Additionally, we explore the integration of piezotronic devices, covering novel piezoelectric materials, their applications in energy harvesting and sensing, and the development of capacitive piezotronic sensors. From an algorithmic perspective, we investigate hardware-aware SNNs by incorporating neuronal device leaky time constants, which significantly impact network performance. This work advances neuromorphic computing for future sensory memory systems by optimizing device technologies and developing hardware-aware algorithms for artificial intelligence and low-power sensing.　❞

SPEAKER
▸　Prof. Chao-Hung WANG (CV*)
▸　Miin Wu School of Computing, National Cheng Kung University

TIME
▸　2025/04/22 (TUE) 13:20 ~ 15:10

VENUE
▸　Classroom 201, Engineergin Building 1