Multi-scopic Cognitive Memory System for Daily Behaviors Recognition

Abstract

With the advancement of artificial intelligence technologies in recent years, research on intelligent robots has progressed. Communication robots that engage psychologically with people are gaining traction. Human behaviors recognition is important in human-robot interaction because it allows robot to comprehend diverse human intention. The catastrophic forgetting problem occurs when traditional deep learning approaches learns new data, so the learned data is forgotten. In this study, we proposed a method called Multi-scopic Cognitive Memory System(MCMS) that mimics the human continuous learning process that can continuously learn new behaviors without forgetting previously learned behaviors. At the same time, we consider selective attention in human behavior recognition for data with individual differences.