Majority of Artificial Neural Network (ANN) implementations in autonomous systems use a fixed/user-prescribed network topology, leading to sub-optimal performance and low portability. The existing neuro-evolution of augmenting topology or NEAT paradigm offers a powerful alternative by allowing the network topology and the connection weights to be simultaneously optimized through an evolutionary process. However, most NEAT implementations allow the consideration of only a single objective. There also persists the question of how to tractably introduce topological diversification that mitigates overfitting to training scenarios. To address these gaps, this paper develops a multi-objective neuro-evolution algorithm. While adopting the basic elements of NEAT, important modifications are made to the selection, speciation, and mutation processes. With the backdrop of small-robot path-planning applications, an experience-gain criterion is derived to encapsulate the amount of diverse local environment encountered by the system. This criterion facilitates the evolution of genes that support exploration, thereby seeking to generalize from a smaller set of mission scenarios than possible with performance maximization alone. The effectiveness of the single-objective (optimizing performance) and the multi-objective (optimizing performance and experience-gain) neuro-evolution approaches are evaluated on two different small-robot cases, with ANNs obtained by the multi-objective optimization observed to provide superior performance in unseen scenarios.
Skip Nav Destination
ASME 2018 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference
August 26–29, 2018
Quebec City, Quebec, Canada
Conference Sponsors:
- Design Engineering Division
- Computers and Information in Engineering Division
ISBN:
978-0-7918-5176-0
PROCEEDINGS PAPER
Multi-Criteria Evolution of Neural Network Topologies: Balancing Experience and Performance in Autonomous Systems
Sharat Chidambaran,
Sharat Chidambaran
University at Buffalo, Buffalo, NY
Search for other works by this author on:
Amir Behjat,
Amir Behjat
University at Buffalo, Buffalo, NY
Search for other works by this author on:
Souma Chowdhury
Souma Chowdhury
University at Buffalo, Buffalo, NY
Search for other works by this author on:
Sharat Chidambaran
University at Buffalo, Buffalo, NY
Amir Behjat
University at Buffalo, Buffalo, NY
Souma Chowdhury
University at Buffalo, Buffalo, NY
Paper No:
DETC2018-86104, V02BT03A039; 14 pages
Published Online:
November 2, 2018
Citation
Chidambaran, S, Behjat, A, & Chowdhury, S. "Multi-Criteria Evolution of Neural Network Topologies: Balancing Experience and Performance in Autonomous Systems." Proceedings of the ASME 2018 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. Volume 2B: 44th Design Automation Conference. Quebec City, Quebec, Canada. August 26–29, 2018. V02BT03A039. ASME. https://doi.org/10.1115/DETC2018-86104
Download citation file:
18
Views
Related Proceedings Papers
Related Articles
A Geometric Path Planner for Car-like Robots
J. Mech. Des (September,2000)
Robot Path Planning in Uncertain Environments: A Language-Measure-Theoretic Approach
J. Dyn. Sys., Meas., Control (March,2015)
Related Chapters
Time-Varying Coefficient Aided MM Scheme
Robot Manipulator Redundancy Resolution
QP Based Encoder Feedback Control
Robot Manipulator Redundancy Resolution
Robot Path Planning Using Wavefront Approach with Virtual Wave HI
International Conference on Advanced Computer Theory and Engineering (ICACTE 2009)