Systems and methods that enable continuous memory-based learning in deep learning and artificial intelligence to continuously run applications across network compute edges

摘要

Lifetime deep neural network (L-DNN) technology revolutionizes deep learning by enabling high-speed post-deployment learning without extensive training, expensive computational resources, or large data storage. It uses the expressive DNN-based subsystem (module A) with the fast learning subsystem (module B) to quickly learn new features without forgetting the previously learned features. Compared to conventional DNNs, L-DNNs use much less data to build a solid network, significantly reduce training time, and learn from devices instead of servers. New knowledge can be added without retraining or storing data. As a result, edge devices with L-DNNs can learn continuously after deployment, and are costly in data collection and labeling, memory and data storage, and computational power. This high-speed local device learning can be used for drone-based inspection of infrastructure and characteristics among security, supply chain monitoring, disaster and emergency response, and other applications.