Evolution of Covariance Functions for Gaussian Process Regression Using Genetic Programming
In this contribution we describe an approach to evolve composite covariance functions for Gaussian processes using genetic programming. A critical aspect of Gaussian processes and similar kernel-based models such as SVM is, that the covariance function should be adapted to the modeled data. Frequently, the squared exponential covariance function is used as a default. However, this can lead to a misspecified model, which does not fit the data well.
In the proposed approach we use a grammar for the composition of covariance functions and genetic programming to search over the space of sentences that can be derived from the grammar.
We tested the proposed approach on synthetic data from two-dimensional test functions, and on the Mauna Loa CO 2 time series. The results show, that our approach is feasible, finding covariance functions that perform much better than a default covariance function. For the CO 2 data set a composite covariance function is found, that matches the performance of a hand-tuned covariance function.
KeywordsGaussian Process Genetic Programming Structure Identification
Unable to display preview. Download preview PDF.
- 2.Duvenaud, D., Lloyd, J.R., Grosse, R., Tenenbaum, J.B., Ghahramani, Z.: Structure Discovery in Nonparametric Regression through Compositional Kernel Search. ArXiv e-prints (February 2013)Google Scholar
- 3.Duvenaud, D., Nickisch, H., Rasmussen, C.E.: Additive Gaussian processes. arXiv preprint arXiv:1112.4394 (2011)Google Scholar
- 4.Gagné, C., Schoenauer, M., Sebag, M., Tomassini, M.: Genetic programming for kernel-based learning with co-evolving subsets selection. In: Runarsson, T.P., Beyer, H.-G., Burke, E.K., Merelo-Guervós, J.J., Whitley, L.D., Yao, X. (eds.) PPSN IX. LNCS, vol. 4193, pp. 1008–1017. Springer, Heidelberg (2006)Google Scholar
- 10.Rasmussen, C.E., Williams, C.K.: Gaussian Processes for Machine Learning. MIT Press (2006)Google Scholar
- 11.Wagner, S.: Heuristic optimization software systems – Modeling of heuristic optimization algorithms in the HeuristicLab software environment. Ph.D. thesis, Institute for Formal Models and Verification, Johannes Kepler University, Linz (2009)Google Scholar
- 12.Wilson, A.G., Prescott Adams, R.: Gaussian Process Covariance Kernels for Pattern Discovery and Extrapolation. ArXiv e-prints (February 2013)Google Scholar