Genetic programming with cross-task knowledge sharing for learning of visual concepts
Inhaltsverzeichnis
Reference
Jaśkowski, W.: Genetic programming with cross-task knowledge sharing for learning of visual concepts. Master’s thesis, Poznan University of Technology, Poznań, Poland (2006),
DOI
Abstract
This thesis concentrates in the fields of computer vision, image understanding and ma- chine learning. We present a novel approach for learning visual concepts from raw image data. Our method is able to learn new concepts and, in result, acquire knowledge that can be than used in standard computer vision tasks such as pattern recognition, object identification or object tracking. The acquired knowledge is encoded in a form of individ- uals/learners which are able to process visual information. The originality of this approach lies also in the fact that individuals do not process raw image data directly. Instead, they operate on a set of attributed visual primitives that are acquired from images in the preliminary stage of the processing. The proposed approach is very general, because it uses the learning principle. For the pur- pose of learning we use genetic programming. Each individual/learner represents a pro- cedure in a form of a tree of operations on sets of visual primitives. The images from the training set in our approach are not labeled in any way and no information is assigned with them. It means that the knowledge is acquired in totally unsupervised way. The only feedback information that states which individual better represents the target concept from the training set is the measure how the individual is able to reconstruct the input image. The rational standing behind this approach is following. We believe that a visual concept can be proved to be acquired and understood only if it can be reconstructed and successfully compared with original image from it was previously learned. In the thesis we present our methodology in details and describe its exemplary im- plementation that is based on the concept of segment as visual primitive. The method is verified on several sets of shapes of simple objects such as triangles, sections, Y let- ters and others. We also elaborate a methodology of cross-task knowledge sharing that is a step towards modularization of knowledge and provide results of massive experiments with cross-task knowledge sharing. Finally, the most important design decisions of software environment that was de- veloped as a platform for computational experiments with our approach are described. A particular emphasis is put on computational performance issues of the developed soft- ware.
Extended Abstract
Bibtex
Used References
[1] Batik svg toolkit. http://xmlgraphics.apache.org/batik/.
[2] Gnu common cpp framework. http://www.gnu.org/software/commoncpp/.
[3] Graph visualization software (graphviz). http://www.graphviz.org/.
[4] Wikipedia – the free encyclopedia. http://wikipedia.org/.
[5] Intel Corporation. Open source computer vision library (opencv), 2000. http://www.intel.com/technology/computing/opencv/.
[6] N.L. Cramer. A representation for the adaptive generation of simple sequential pro- grams. In Proceedings of the 1st International Conference on Genetic Algorithms, pages 183–187, Mahwah, NJ, USA, 1985. Lawrence Erlbaum Associates, Inc.
[7] J.H. Holland. Adaptation in Natural and Artificial Systems: An Introductory Anal- ysis with Applications to Biology, Control, and Artificial Intelligence. University of Michigan Press, Ann Arbor, MI, USA, 1975.
[8] D. Howe. The free on-lline dictionary of computing. http://foldoc.org.
[9] Sun Microsystems Inc. Java advanced imaging api specification, 2001. Version 1.2.
[10] J.R. Koza. Hierarchical genetic algorithms operating on populations of computer programs. In Proceedings of the 11th International Joint Conference on Artificial In- telligence, volume I, pages 768–774, San Mateo, CA, USA, 1989. Morgan Kaufmann.
[11] J.R. Koza. Genetic Programming: On the Programming of Computers by Means of Natural Selection. The MIT Press, Cambridge, MA, USA, 1992.
[12] J.R. Koza. Genetic Programming II: Automatic Discovery of Reusable Programs. The MIT Press, Cambridge, MA, USA, 1994.
[13] J.R. Koza, F.H. Bennett III, D. Andre, and M.A. Keane. Genetic Programming III: Darwinian Invention and Problem Solving. Morgan Kaufmann Publishers, San Francisco, CA, USA, 1999.
[14] J.R. Koza, M.A. Keane, M.J. Streeter, W. Mydlowec, J. Yu, and G. Lanza. Genetic Programming IV. Routine Human-Competitive Machine Intelligence.
[15] K. Krawiec. Pairwise comparison of hypotheses in evolutionary learning. In C.E. Brodley and A. Pohoreckyj Danyluk, editors, Proceedings of the Eighteenth Interna- tional Conference on Machine Learning, pages 266–273, San Francisco, USA, 2001. Morgan Kaufmann.
[16] K. Krawiec. Visual learning by coevolutionary feature synthesis. IEEE Trans. on Systems, Man, and Cybernetics, Part B: Cybernetics, 35(3):409–425, June 2005.
[17] K. Krawiec. Learning high-level visual concepts using attributed primitives and ge- netic programming. In Proceedings Lecture Notes in Computing Science, volume 3907, pages 515–519, Berlin/Heidelberg, April 2006. Springer.
[18] K. Krawiec and B. Bhanu. Visual learning by evolutionary feature synthesis. In Mishra N. Fawcett, T., editor, Proceedings of the Twentieth International Conference on Machine Learning (ICML 2003), pages 376–383, Menlo Park, CA, USA, 2003. AAAI Press.
[19] S. Luke. Ecj – evolutionary computation system, 2002. http://cs.gmu.edu/eclab/projects/ecj/.
[20] D. Marr. Early processing of visual information. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 275(942):483–519, 1976.
[21] J. Peng and B. Bhanu. Closed-loop object recognition using reinforcement learn- ing. IEEE Transactions on Pattern Analysis and Machine Intelligence, (20):139–154, 1998a.
[22] J. Peng and B. Bhanu. Delayed reinforcement learning for adaptive image segmenta- tion and feature extraction. IEEE Transactions on Systems, Man and Cybernetics, (28):282–488, August 1998b.
[23] M. Picardi and T. Jan. Recent advances in computer vision. The Industrial Physicist, 9(1):18–21, February/March 2003.
[24] J. Segen. Gest: A learning computer vision system that recognizes hand gestures. In G. Tecuci, editor, Machine Learning. A Multistrategy Approach.
[25] S.F. Smith. A learning system based on genetic adaptive algorithms. PhD thesis, 1980.
[26] A. Teller and M. Veloso. A constrolled experiment: evolution for learning difficult image classification. Lecture Notes in Computer Science, (990):165–185, 1995.
[27] B. Wieloch. Genetic programming with knowledge modularization for learning of visual concepts. Master’s thesis, Poznań University of Technology, 2006.
[28] X. Yao, editor. Evolutionary computation. Theory and applications.
Links
Full Text
http://www.cs.put.poznan.pl/wjaskowski/pub/papers/jaskowski06crosstask.pdf
