Cognitive-based search strategies for complex bionanotechnology design derived through symbiotic human and agent-based approaches

Paul F. Egan, Jonathan Cagan, Christian Schunn, Philip R. Leduc

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

10 Scopus citations

Abstract

Complex systems are challenging for engineers to understand and design. This work demonstrates a synergistic cognitive and agentbased methodology for developing and implementing rule-based strategies that improve human search performance in optimization design tasks. The domain of our study is the design of synthetic myosin-based systems, the biologically-based building block of muscle. We began with an initial cognitive study of users solving design tasks with three varied difficulties using a graphical user interface, and tracked how they manipulated design variables in their search process. User search behaviors resulting in the best and worst designs were then examined. Trends were identified that were used to formulate three strategies automated by computational agents solving the same tasks as the users. The most successful identified strategy implemented by the agents was a combination of univariate searches to learn parameter relationships and then applying that knowledge in greedy local searches. On one of the three tasks, an initial random search improved results. A subsequent cognitive study was conducted with users implementing the best agent-tested strategies. Users implementing the strategy performed significantly better than users performed in the first study with no provided strategy. These results show the power of synergistic human and agent-based approaches, in that cognitive-based findings can provide a starting place for computational search algorithms to begin testing strategies. Experimentation through agent-based methods via fast and extensive automated searches can then produce effective strategies that are given back to users. Our primary findings demonstrate that these agenttested strategies significantly improve human search performance in designing these complex systems.

Original languageEnglish
Title of host publication2nd Biennial International Conference on Dynamics for Design; 26th International Conference on Design Theory and Methodology
PublisherAmerican Society of Mechanical Engineers (ASME)
ISBN (Electronic)9780791846407
DOIs
StatePublished - 2014
EventASME 2014 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, IDETC/CIE 2014 - Buffalo, United States
Duration: Aug 17 2014Aug 20 2014

Publication series

NameProceedings of the ASME Design Engineering Technical Conference
Volume7

Conference

ConferenceASME 2014 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, IDETC/CIE 2014
CountryUnited States
CityBuffalo
Period08/17/1408/20/14

Fingerprint Dive into the research topics of 'Cognitive-based search strategies for complex bionanotechnology design derived through symbiotic human and agent-based approaches'. Together they form a unique fingerprint.

Cite this