Abstract
Feature selection is a task of great importance. Many feature selection methods have been proposed, and can be divided generally into two groups based on their dependence on the learning algorithm/classifier. Recently, a feature selection method that selects features at the same time as it evolves neural networks that use those features as inputs called Feature Selective NeuroEvolution of Augmenting Topologies (FS-NEAT) was proposed by Whiteson et al. In this paper, a novel feature selection method called Feature Deselective NeuroEvolution of Augmenting Topologies (FD-NEAT) is presented. FD-NEAT begins with fully connected inputs in its networks, and drops irrelevant or redundant inputs as evolution progresses. Herein, the performances of FD-NEAT, FS-NEAT and traditional NEAT are compared in some mathematical problems, and in a challenging race car simulator domain (RARS). On the whole, the results show that FD-NEAT significantly outperforms FS-NEAT in terms of network performance and feature selection, and evolves networks that offer the best compromise between network size and performance.
Similar content being viewed by others
References
Zongker D, Jain A (1996) Algorithms for feature selection: an evaluation. In: Proceedings of the 13th international conference on pattern recognition. Vienna, Austria, pp 18–22
Kittler J (1978) Feature set search algorithms. In: Chen CH (ed) Pattern recognition and signal processing. Sijthoff and Noordhoff, Alphen aan den Rijn, Netherlands, pp 41–60
Mao KZ (2002) Fast orthogonal forward selection algorithm for feature subset selection. IEEE Trans Neural Netw 13:1218–1224
Narendra PM, Fukunaga K (1977) A branch and bound algorithm for feature subset selection. IEEE Trans Comput c-26:917–922
Ferri FJ, Pudil P, Hatef M, Kittler J (1994) Comparative study of techniques for large-scale feature selection. In: Gelsema ES, Kanal LN (eds) Pattern recognition in practice IV. Elsevier Science B.V., Amsterdam, pp 403–413
Pudil P, Novovicova J, Kittler J (1994) Floating search methods in feature selection. Pattern Recognit Lett 15:1119–1125
Jain A, Zongker D (1997) Feature selection: evaluation, application, and small sample performance. IEEE Trans Pattern Anal Machine Intell 19:153–157
Kudo M, Sklansky J (2000) Comparison of algorithms that select features for pattern classifiers. Pattern Recognit 33:25–41
Kohavi R, John GH (1997) Wrappers for feature subset selection. Artif Intell 97:273–324
Langley P (1994) Selection of relevant features in machine learning. In: Proceedings of the AAAI fall symposium on relevance. AAAI Press, New Orleans
Bonnlander BV, Weigend AS (1994) Selecting input variables using mutual information and nonparametric density estimation. In: Proceedings of the 1994 international symposium on artificial neural networks (ISANN’94). Tainan, Taiwan, pp 42–50
Yang J, Honavar V (1998) Feature subset selection using a genetic algorithm. IEEE Intell Syst 13:44–49
Whiteson S, Stanley KO, Miikkulainen R (2004) Automatic feature selection in neuroevolution. In: Proceedings of the genetic and evolutionary computation conference (GECCO), Seattle, Washington, USA
Whiteson S, Stone P, Stanley KO, Miikkulainen R, Kohl N (2005) Automatic feature selection in neuroevolution. In: Proceedings of the genetic and evolutionary computation conference (GECCO), Washington, DC, USA
Stanley KO, Miikkulainen R (2002) Evolving neural networks through augmenting topologies. Evol Comput 10:99–127
Stanley KO, Miikkulainen R (2004) Competitive coevolution through evolutionary complexification. J Artif Intell Res 21:63–100
Langley P, Sage S (1994) Oblivious decision trees and abstract cases. In: Working notes of the AAAI-94 workshop on case-based reasoning. AAAI Press, Seattle, pp 113–117
Kelly JDK, Davis L (1991) Hybridizing the genetic algorithm and the k nearest neighbors classification algorithm. In: Belew RK, Booker LB (eds) Proceedings of the 4th international conference on genetic algorithms. Morgan Kaufmann, San Diego, pp 377–383
Timin ME (1995) The robot auto racing simulator. Available via http://rars.sourceforge.net
Gomez F, Miikkulainen R (1998) 2-D pole balancing with recurrent evolutionary networks. In: Proceedings of the international conference on artificial neural networks (ICANN-98), Skovde, Sweden. Elsevier, New York
Gomez F, Miikkulainen R (1997) Incremental evolution of complex general behavior. Adapt Behav 5:317–342
Gruau F, Whitley D, Pyeatt L (1996) A comparison between cellular encoding and direct encoding for genetic neural networks. In: Koza JR, Goldberg DE, Fogel DB, Riolo, RL (eds) Proceedings of the first annual conference on genetic programming, Cambridge, MA, pp 81–89
Stanley KO (2004) Efficient evolution of neural networks through complexification. PhD thesis, The University of Texas at Austin
Harvey I (1992) Species adaptation genetic algorithms: a basis for a continuing SAGA. In: Varela FJ, Bourgine P (eds) Proceedings of the 1st European conference on artificial life, toward a practice of autonomous systems. MIT Press/Bradford Books, Cambridge, pp 346–354
Cliff D, Harvey I, Husbands P (1992) Incremental evolution of neural network architectures for adaptive behaviour. Technical report CSRP256, School of Cognitive and Computing Sciences, University of Sussex, UK
Gomez FJ, Miikkulainen R (1999) Solving non-markovian control tasks with neuroevolution. In: Proceedings of the international joint conference on artificial intelligence. Stockholm, Sweden. Morgan Kaufmann, Denver
Saravanan N, Fogel DB (1995) Evolving neural control systems. IEEE Expert 10(3):23–27
Wieland AP (1991) Evolving neural network controllers for unstable systems. In: Proceedings of the international joint conference on neural networks, Seattle, WA. Piscataway, New Jersey, pp 667–673
Yao X (1999) Evolving artificial neural networks. Proc IEEE 87:1423–1447
Moriarty DE, Miikkulainen R (1996) Efficient reinforcement learning through symbiotic evolution. Machine Learn 22:11–32
Radcliffe NJ (1993) Genetic set recombination and its application to neural network topology optimization. Neural Comput Appl 1:67–90
Dasgupta D, McGregor D (1992) Designing application-specific neural networks using the structured genetic algorithm. In: Proceedings of the international conference on combinations of genetic algorithms and neural networks. IEEE Computer Society Press, USA, pp 87–96
Pujol JCF, Poli R (1997) Evolution of the topology and the weights of neural networks using genetic programming with a dual representation. Technical report CSRP-97-7, School of Computer Science, The University of Birmingham, Birmingham B15 2TT, UK
Gruau F (1993) Genetic synthesis of modular neural networks. In: S. Forrest (ed) Proceedings of the fifth international conference on genetic algorithms. Morgan Kaufmann, San Mateo, CA, pp 318–325
Angeline PJ, Saunders GM, Pollack JB (1993) An evolutionary algorithm that constructs recurrent neural networks. IEEE Trans Neural Netw 5:54–65
Wieland A (1991) Evolving neural network controllers for unstable systems. In: Proceedings of the international joint conference on neural networks. Piscataway, New Jersey, pp 667–673
Stanley KO, Bryant BD, Miikkulainen R (2005) Real-time neuroevolution in the NERO video game. IEEE Trans Evol Comput 9:653–668
Stanley KO, Miikkulainen R (2004) Evolving a roving eye for go. In: Proceedings of the genetic and evolutionary computation conference (GECCO). Springer, New York, pp 1226–1238
Schlessinger E, Bentley PJ, Lotto RB (2005) Analysing the evolvability of neural network agents through structural mutations. In: Capcarrere M (ed) Proceedings of the European conference on artificial life (ECAL 2005). Springer, Berlin, pp 312–321
Yao X, Liu Y (1996) Towards designing artificial neural networks by evolution. Appl Math Comput 91:83–90
Kohl N, Stanley K, Miikkulainen R, Samples M, Sherony R (2006) Evolving a real-world vehicle warning system. In: Proceedings of the genetic and evolutionary computation conference (GECCO), Seattle, Washington, USA, pp 1681–1688
Gomez F, Miikkulainen R (1998) 2-D pole balancing with recurrent evolutionary networks. In: Proceedings of the international conference on artificial neural networks (ICANN). Skovde, Sweden. Elsevier, New York
Goldberg DE, Richardson J (1987) Genetic algorithms with sharing for multimodal function optimization. In: Proceedings of the second international conference on genetic algorithms. Lawrence Erlbaum Associates, Hillsdale, pp 41–49
Gruau F, Whitley D, Pyeatt L (1996) A comparison between cellular encoding and direct encoding for genetic neural networks. Technical Report NC-TR-96-048, NeuroCOLT
Gomez F, Miikkulainen R (1999) Solving non-Markovian control tasks with neuroevolution. In: Proceedings of the 16th international joint conference on artificial intelligence. Morgan Kaufmann, Denver
Coons KE, Robatmili B, Taylor ME, Maher BA, Burger D, McKinley KS (2008) Feature selection and policy optimization for distributed instruction placement using reinforcement learning. In: Proceedings of the 7th international joint conference on parallel architectures and compilation techniques (PACT), Toronto, Ontario, Canada
Skalak DB (1994) Prototype and feature selection by sampling and random mutation hill-climbing algorithms. In: Proceedings of the 11th international conference on machine learning. Morgan Kaufmann, New Brunswick, pp 293–301
Mayr C (2003) NEAT Matlab. Available via http://www.cs.utexas.edu/~nn/soft-view.php?SoftID=23. Accessed 4 Sept 2008
Ethembabaoglu A, Whiteson S (2008) Automatic feature selection using FS-NEAT. Technical report IAS-UVA-08-02, Intelligent Autonomous Systems Group, University of Amsterdam
Author information
Authors and Affiliations
Corresponding author
Appendix 1
Appendix 1
1.1 Common parameters for regular NEAT, FS-NEAT and FD-NEAT
The common parameters for regular NEAT, FS-NEAT and FD-NEAT, their definitions and values used in the RARS experiments are displayed in Table 2. The stagnation parameters to measure stagnation in species fitness were disabled in the experiments so that species would not die out.
1.2 Differing parameters
There are two parameters that differ between FS-NEAT, FD-NEAT and regular NEAT in the RARS experiments. These are (1) the probability of a connection gene being re-enabled in the offspring if it was inherited disabled (Gene Re-enable Prob.), and (2) the probability of a connection gene being disabled in the offspring if it was inherited enabled (Gene Disable Prob.). Gene Disable Prob. only acts on the input connections, and is only applicable to FD-NEAT. Gene Re-enable Prob. applies to all the connection genes, and is not only focused on the input connections. The values of each parameter for FS-NEAT, FD-NEAT and regular NEAT are displayed in Table 3. The values of the two parameters for FD-NEAT were obtained experimentally, namely, experimentations showed that FD-NEAT performed best when these values were used in the algorithm.
Rights and permissions
About this article
Cite this article
Tan, M., Hartley, M., Bister, M. et al. Automated feature selection in neuroevolution. Evol. Intel. 1, 271–292 (2009). https://doi.org/10.1007/s12065-009-0018-z
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12065-009-0018-z