Real-time human-guided evolutionary optimization for material design

From Anita Borg Institute Wiki

Jump to: navigation, search

Real-time human-guided evolutionary optimization for material design.jpg

Presenter: Elena Erbiceanu, University of Central Florida, The Media Convergence Laboratory

Co-Authors: Mark Colbert, Kenneth O. Stanley, Charles E. Hughes


Many artistic tasks in creative domains, such as content generation in the movie and entertainment industries, rely on extensive human-processing, although most of the process could be automated. We introduce “real-time human-guided evolutionary optimization” as a solution to lessen the artist’s effort. Illustrating this, “Surface Highlighting via Interactive NeuroEvolution” defines visual appearance of objects in a scene with only sparse guidance by the artist. Unlike traditional approaches, evolution happens interactively, in real-time. Moreover, it transparently “suggests” creative solutions under the artist’s guidance. We discuss major challenges in designing a real-time creative evolutionary algorithm that instantaneously adapts to sparse, incremental user-input.

More details: http://mcl.ucf.edu/research/neuroimd/

Video demonstrating how it works: http://mcl.ucf.edu/research/neuroimd/material_design.avi

Personal tools