Faculty Publications and Other Works -- Ecology and Evolutionary Biology

Source Publication

BioInspired Engineering

Author ORCID Identifier

https://orcid.org/0000-0003-0233-1404

Document Type

Book Chapter

Publication Date

2019

Abstract

Nature-Inspired algorithms have given us elegant solutions across the a broad spectrum of design challenges[1]. Efforts within cyber-systems are equally replete with cases in which algorithms either adapted from, or at least inspired by, observations of natural biological systems, ranging from efficient packet routing [2], to anomaly detection [3], and beyond [4]. It is tempting to turn to nature for many reasons, both aesthetic and practical. At the most basic of practical reasons, these algorithms are already-proven techniques, avoiding the potential for catastrophic failure (so long as the application is sufficiently similar to the natural setting). Of course, simple assurance of sufficient function is very different from assurance of locally maximized, or globally optimal performance, and (as we will discuss in detail below) it is rare for a natural system to truly optimize algorithms in ways that allow for straightforward adaptation outside their native environment. Natural systems can also function as ‘black boxes’ where design effort can be bypassed by demonstrable success. Of course, black box solutions are not without their own requisite investment of effort. To be assured that algorithms from natural settings function adequately in humandesigned settings, the analogy between the settings must be sufficiently robust, including likely bounds on constraints in both the behaviors of the system’s components and system-wide acceptable outcomes. Constructing these analogies is often itself time-consuming and it may be impossible to understand (at least initially) how any departure from natural algorithms may affect performance outcomes. Ideally, exploitation of systems that are initially treated as black box solutions can lead to discovery via fruitful interrogation into why a seemingly unlikely solution actually works, then allowing purposeful design that leverages that insight. In this chapter, we will discuss how natural systems achieve robust and efficient solutions and the implications this ontology has for helping inform how and when it may be most fruitful to exploit them, and when instead we should probably forgo analogy and instead build tailored tools from scratch. Further, we will consider how it might be possible to move beyond this case-by-case analogy paradigm to exploit the deeper design elements that make natural systems successful at converging on sufficient solutions and then employ these more fundamental tools in purposeful design.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS