Cooperative Scheduling Schemes for Explainable DNN Acceleration in Satellite Image Analysis and Retraining

The deep learning-based satellite image analysis and retraining systems are getting emerging technologies to enhance the capability of sophisticated terrestrial objects. In principle, apply explainable DNN model for process retraining, we consider a new acceleration scheduling mechanism. Especially, conventional schemes cause serious performance degradation due computational complexity costs in retraining. this article, overcome degradation, propose cooperative process. For purpose it, define latency energy cost modeling derive optimized processing time required acceleration. show minimum considered proposed via layer-level management on FPGA-GPU system. addition, evaluate using an adaptive unlabeled data selection scheme with confidence threshold semi-supervised learning driven parallelism accelerating experimental results demonstrate that reduce by up about 40% while guaranteeing constraints.