Towards a Design Pattern for Adaptive Systems Inspired by the Neocortex

Abstract

This paper discusses a general design pattern that can provide systems engineering guidance for building adaptive systems. The Neocortex Adaptive System Pattern (NASP) architecture is an adaptive decision‐making architecture. It is derived from the physical architecture observed within the neocortex. It allows different adaptive system components with diverse methodologies and techniques to coexist and cooperate within a single system. A literature review of relevant neuroscience literature is provided. An architectural view, along with a taxonomy of architecture features, is explained in detail. Experiment results that compare an implemented NASP system, a non‐NASP adaptive system, and rule‐based systems within a combative simulation are presented. The NASP design pattern unifies the diverse set of algorithms and technologies found across the spectrum of adaptive systems. Evidence from this study indicates that multiple methods and time windows for adaptation can result in superior performance of the overall system. The implications of this result infer a clear preference for Bayesian‐based adaptive systems over neural network–based adaptive systems when the system faces time‐sensitive problems. This paper supports the claim of a design pattern by providing an example system that shares the same architectural features as NASP.

Researching Adaptive Systems with RRobots

“…This enabled the creation of many types of adaptive systems in the form of robot combatants. The results of this effort are documented in Phillips and Blackburn 13,18 . A self-healing adaptive systems study required the addition of new agents and agent capabilities.…”

“…An adaptive robot may use probability models to select behaviors based on an incoming stimulus event. Such a system may change its behavior set to maximize a win probability 13 . The environment was changed so that it collects additional information about each tanks behavior.…”