A5041194022- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Neuroscience and Neural Engineering
- CCD and CMOS Imaging Sensors
- Video Coding and Compression Technologies
- Image and Video Quality Assessment
- Caching and Content Delivery
University of South Alabama
2022-2024
Deep Neural Networks (DNNs) are widely used in edge AI. But the complex perception and decision-making demands overlarge computation makes DNN architecture very sophisticated. Memristors have multilevel resistance property that enables faster in-memory to remove bottleneck caused by von Neumann CMOS technology. However, Stuck-at-Fault (SAF) defect of memristor generated from immature fabrication heavy device utilization memristor-based AI commercially unavailable. To mitigate this problem,...
Stuck-At-Fault (SAF) defect of memristor generated from immature fabrication and heavy device utilization makes neuromorphic computing systems commercially unavailable. To mitigate this problem, a Reconfigurable Mapping Algorithm (RMA) is proposed in paper. Based on the analysis for VGG8 model with CIFAR10 dataset, experiment results show that RMA efficient restoring inference accuracy up to 90% (the original without SAF) under SAFs 0.1% 50%, where Stuck-At-One (SA1): Stuck-At-Zero (SA0) =...
The demand for mobile video streams is constantly increasing. With this comes a need devices to receive more videos at ever increasing quality. However, due the large size of data and intensive computational requirements, streaming requires frequent memory access that consume substantial amount device power; as result, battery life limited. In paper, we present content-adaptable Region-of-Interest (ROI)-aware storage technique promotes power savings. During encoding process on transmitting...