A self-supervised learning system for category detection by integrating information from several sensors
Kouichirou Yamauchi, Mikiya Oota, Naohiro Ishii · 2002
An artificial neural network is a useful tool for pattern recognition because the network can realize nonlinear mapping between input and output space. This ability is tuned by supervised learning methods such as backpropagation. In supervised learning methods, desired outputs of the neural network are needed. However, the desired outputs are usually unknown in unpredictable environments. To solve this problem, the paper presents a self-supervised learning system for autonomous knowledge acquisition. This system learns categories of objects and boundaries between the categories automatically by integrating information from several sensors. We assume that patterns of these sensory inputs are always ambiguous and include noise according to deformation of the objects.