Hierarchical Temporal Memory Examined

Abstract

During the summer of 2016 the author completed a ten-week internship at IBM’s Almaden Research facility, which constituted as the research aspect of this SIP. She assisted Dr. Geoffrey Burr in his research on Machine Intelligence by improving a Neural Network project that IBM developed using an article by Dr. Jeff Hawkins on Machine Intelligence, Hierarchical Temporal Memory (HTM). The first portion of this paper contains an explanation of the basic concepts behind Machine Intelligence and Neural Networks, followed by a description of Dr. Hawkins’ paper as it relates to her work. The report concludes by describing two of the tasks worked on during the internship at IBM: the creation of measuring tools to monitor the amount of memory space the project requires and applying the HTM design to text application problems.