Human-guided search

  • Authors:
  • Gunnar W. Klau;Neal Lesh;Joe Marks;Michael Mitzenmacher

  • Affiliations:
  • CWI, Amsterdam, Netherlands 1090;Dimagi, Inc., Cambridge, USA 02139;Walt Disney Animation Studios, Burbank, USA 91521;Harvard School of Engineering and Applied Sciences, Cambridge, USA 02138

  • Venue:
  • Journal of Heuristics
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present a survey of techniques and results from the Human-Guided Search (HuGS) project, an effort to investigate interactive optimization. HuGS provides simple and general visual metaphors relating to local search operations that allow users to guide the exploration of the search space. These metaphors apply to a wide variety of problems and combinatorial optimization algorithms, which we demonstrate by describing the HuGS toolkit and as well as eight diverse applications we developed using it. User experiments show that human guidance can improve the performance of powerful heuristic search algorithms. HuGS is also a valuable development environment for understanding and improving optimization algorithms. Although HuGS was designed for human-computer interaction, for two different problems we have used the HuGS code base to develop completely automatic heuristic algorithms that produced at the time new best automatic results on benchmark problem instances.