Experiments with Genetic Programming for Active Vision Tasks

Created by W.Langdon from gp-bibliography.bib Revision:1.4192

  title =        "Experiments with Genetic Programming for Active Vision
  author =       "Jay F. Winkeler and B. S. Manjunath",
  howpublished = "CiteSeer",
  year =         "1997",
  keywords =     "genetic algorithms, genetic programming",
  annote =       "The Pennsylvania State University CiteSeer Archives",
  language =     "en",
  oai =          "oai:CiteSeerPSU:160647",
  rights =       "unrestricted",
  URL =          "http://citeseer.ist.psu.edu/cache/papers/cs/1845/http:zSzzSzvivaldi.ece.ucsb.eduzSzuserszSzjayzSzPAPERSzSzucsb_97_19.pdf/experiments-with-genetic-programming.pdf",
  URL =          "http://citeseer.ist.psu.edu/160647.html",
  size =         "16 pages",
  abstract =     "This paper examines genetic programming as a machine
                 learning technique for active vision. Genetic
                 programming can be used as a reinforcement learning
                 technique to learn solutions to complex problems based
                 solely on performance at the overall task. Solutions
                 for camera calibration and target tracking are learned
                 by this method. The camera calibration solutions are
                 marginally better than second order least-squares
                 approximations. A solution to the task of tracking a
                 moving target is learned without requiring camera
                 calibration or motion detection to be solved
                 separately. The human effort required to achieve these
                 solutions remains nearly constant despite the
                 increasing difficulty of the tasks.",

Genetic Programming entries for Jay F Winkeler B S Manjunath