The HTM Learning Algorithm
Kjell Jorgen Hole · 2016
According to the fail fast principle in Chap. 4 , we need to learn from systems’ abnormal behavior and downright failures to achieve anti-fragility to classes of negative events. The earlier we can detect problems, the smaller the negative consequences are and the faster we can start learning how to improve the systems. Since humans are not good at detecting anomalies, especially in streaming data from large cloud applications, a form of automatic anomaly detection is needed. This first chapter of Part IV introduces a general learning algorithm based on Jeff Hawkins’s developing theory of how the brain learns, called hierarchical temporal memory (HTM). The HTM learning algorithm is used in the next chapter to detect anomalies in a system’s behavior. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.