OPEN ACCESS
Predictive learning is a machine learning approach where an agent tries to create a model of its environment by trying out various actions under variable circumstances. The agent eventually creates planning operators through awareness of its actions' effects, enabling it to act purposefully in the world. This learning method allows the agent to learn with minimal pre-existing mental structure. Two modules comprise the basic architecture of a predictive learning model: a sensorimotor system (lower box), which engages with the environment, and a predictor (upper box), which mimics the sensorimotor system in the brain.