Modelling HTM Learning and Prediction for Robotic Path-Learning

摘要

Various machine learning models have so far been used for training robots to perform different tasks in the context of Industry 4.0. However, following the advances in neuroscience, new models are being pursued which are biologically inspired. One such model is the Hierarchical Temporal Memory (HTM) which models a neural network by drawing inspirations from human neocortex. This model is however a theoretical one, though its performance in multiple scenarios is worth taking note of. In this paper, the authors model the deviation in learning for HTM when applied to a robotic path learning scenario and investigated different parameters which influence the learning.