Revisando Redes Bayesianas através da Introdução de Variáveis Não-observadas
Resumo
Uma questão importante em aprendizado de redes Bayesianas (RB) é como aprender a estrutura da rede na presença de variáveis não-observadas. Neste artigo, propomos uma nova abordagem que utiliza técnicas de revisão de teoria. O algoritmo proposto, denominado DAHVI, aplica uma heurística discriminativa e a partir dos exemplos identifica pontos potenciais na RB à inclusão de uma variável não-observada. Essas variáveis são incluídas através de um operador de revisão proposto neste artigo. O DAHVI é capaz de introduzir tantas variáveis não-observadas quantas forem necessárias para encontrar uma RB com uma melhor avaliação, podendo ser aplicado também a RB esparsas. Avaliamos com sucesso o DAHVI em 12 datasets reais.Referências
Binder, J., Koller, D., Russell, S., and Kanazawa, K. (1997). Adaptive probabilistic networks with hidden variables. Machine Learning, 29:213–244.
Buntine, W. (1991). Theory refinement on bayesian networks. In Proceedings Seventeenth Conference Uncertainty in Artificial Intelligence, pages 52–60, San Mateo, CA.
Dempster, A. P., Laird, N. M., and Rubin, D. B. (1977). Maximum likelihood from incomplete data via the EM algorithm. Royal Stat Soc, 39:1–39.
Elidan, G. and Friedman, N. (2005). Learning hidden variable networks: The information bottleneck approach. Journal of Machine Learning Research, 6:81–127.
Elidan, G., Lotner, N., Friedman, N., and Koller, D. (2000). Discovering hidden variables: a structure-based approach. In Neural Information Processing Systems, volume 13, pages 479–485.
Friedman, N. (1998). The bayesian structural EM algorithm. In UAI, pages 129–138.
Friedman, N., Geiger, D., and Goldszmidt, M. (1997). Bayesian network classifiers. Machine Learning, 29:131–163.
Grossman, D. and Domingos, P. (2004). Learning bayesian network classifiers by maximizing conditional likelihood. In Proc. 21th Int. Conference on Machine Learning, pages 361–368.
Heckerman, D. (1995). A tutorial on learning bayesian networks. Technical report, Microsoft Research.
Kohavi, R. (1995). A study of cross-validation and bootstrap for accuracy estimation and model selection. In Proceedings of the International Joint Conference on Artificial Intelligence(IJCAI), pages 1137–1145.
Lauritzen, S. L. (1995). The em algorithm for graphical association models with missing data. Comp. Stat.and Data Ana., 19:191–201.
Nadeau, C. and Bengio, Y. (2003). Inference for the generalization error. Machine Learning, 52(3):239–281.
Ramachandran, S. and Mooney, R. (1998). Theory refinement of bayesian networks with hidden variables. In Proc. 15th Int. Conference on Machine Learning, pages 454–462.
Wrobel, S. (1996). First-order theory refinement. Advances in Inductive Logic Programming, pages 14–33.
Buntine, W. (1991). Theory refinement on bayesian networks. In Proceedings Seventeenth Conference Uncertainty in Artificial Intelligence, pages 52–60, San Mateo, CA.
Dempster, A. P., Laird, N. M., and Rubin, D. B. (1977). Maximum likelihood from incomplete data via the EM algorithm. Royal Stat Soc, 39:1–39.
Elidan, G. and Friedman, N. (2005). Learning hidden variable networks: The information bottleneck approach. Journal of Machine Learning Research, 6:81–127.
Elidan, G., Lotner, N., Friedman, N., and Koller, D. (2000). Discovering hidden variables: a structure-based approach. In Neural Information Processing Systems, volume 13, pages 479–485.
Friedman, N. (1998). The bayesian structural EM algorithm. In UAI, pages 129–138.
Friedman, N., Geiger, D., and Goldszmidt, M. (1997). Bayesian network classifiers. Machine Learning, 29:131–163.
Grossman, D. and Domingos, P. (2004). Learning bayesian network classifiers by maximizing conditional likelihood. In Proc. 21th Int. Conference on Machine Learning, pages 361–368.
Heckerman, D. (1995). A tutorial on learning bayesian networks. Technical report, Microsoft Research.
Kohavi, R. (1995). A study of cross-validation and bootstrap for accuracy estimation and model selection. In Proceedings of the International Joint Conference on Artificial Intelligence(IJCAI), pages 1137–1145.
Lauritzen, S. L. (1995). The em algorithm for graphical association models with missing data. Comp. Stat.and Data Ana., 19:191–201.
Nadeau, C. and Bengio, Y. (2003). Inference for the generalization error. Machine Learning, 52(3):239–281.
Ramachandran, S. and Mooney, R. (1998). Theory refinement of bayesian networks with hidden variables. In Proc. 15th Int. Conference on Machine Learning, pages 454–462.
Wrobel, S. (1996). First-order theory refinement. Advances in Inductive Logic Programming, pages 14–33.
Publicado
20/07/2009
Como Citar
REVOREDO, Kate; PAES, Aline; ZAVERUCHA, Gerson; COSTA, Vitor Santos.
Revisando Redes Bayesianas através da Introdução de Variáveis Não-observadas. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 7. , 2009, Bento Gonçalves/RS.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2009
.
p. 422-431.
ISSN 2763-9061.
