首页>
外国专利>
METHOD AND DEVICE FOR ON-DEVICE CONTINUAL LEARNING OF A NEURAL NETWORK WHICH ANALYZES INPUT DATA, AND METHOD AND DEVICE FOR TESTING THE NEURAL NETWORK TO BE USED FOR SMARTPHONES, DRONES, VESSELS, OR MILITARY PURPOSE
METHOD AND DEVICE FOR ON-DEVICE CONTINUAL LEARNING OF A NEURAL NETWORK WHICH ANALYZES INPUT DATA, AND METHOD AND DEVICE FOR TESTING THE NEURAL NETWORK TO BE USED FOR SMARTPHONES, DRONES, VESSELS, OR MILITARY PURPOSE
In a method for on-device continual learning of a neural network that analyzes input data for use in smartphones, drones, ships, or military purposes, the learning device (a) samples new data and performs a preset first volume, causing the previously trained original data generator network to pre-synthesis data corresponding to a k-dimensional random vector, wherein the first pre-synthesis data corresponds to previous data used to previously train the original data generator network. repeating the process of outputting , so that the pre-synthesis data becomes a second volume, and generating a batch to be used for current learning; and (b) causing the neural network to generate output information corresponding to the arrangement. The method may be performed by a generative adversarial network (GAN), online learning, or the like, and has the effect of saving resources such as storage, preventing catastrophic forgetting, and protecting personal information.
展开▼