Machine Learning

, Volume 86, Issue 2, pp 233–272 | Cite as

Learning by extrapolation from marginal to full-multivariate probability distributions: decreasingly naive Bayesian classification

  • Geoffrey I. Webb
  • Janice R. Boughton
  • Fei Zheng
  • Kai Ming Ting
  • Houssam Salem


Averaged n-Dependence Estimators (AnDE) is an approach to probabilistic classification learning that learns by extrapolation from marginal to full-multivariate probability distributions. It utilizes a single parameter that transforms the approach between a low-variance high-bias learner (Naive Bayes) and a high-variance low-bias learner with Bayes optimal asymptotic error. It extends the underlying strategy of Averaged One-Dependence Estimators (AODE), which relaxes the Naive Bayes independence assumption while retaining many of Naive Bayes’ desirable computational and theoretical properties. AnDE further relaxes the independence assumption by generalizing AODE to higher-levels of dependence. Extensive experimental evaluation shows that the bias-variance trade-off for Averaged 2-Dependence Estimators results in strong predictive accuracy over a wide range of data sets. It has training time linear with respect to the number of examples, learns in a single pass through the training data, supports incremental learning, handles directly missing values, and is robust in the face of noise. Beyond the practical utility of its lower-dimensional variants, AnDE is of interest in that it demonstrates that it is possible to create low-bias high-variance generative learners and suggests strategies for developing even more powerful classifiers.


Bayesian learning Classification learning Probabilistic learning Averaged one-dependence estimators Naive Bayes Semi-naive Bayesian learning Learning without model selection Ensemble learning Feating 


  1. Affendey, L., Paris, I., Mustapha, N., Sulaiman, M., & Muda, Z. (2010). Ranking of influencing factors in predicting students’ academic performance. Information Technology Journal, 9(4), 832–837. CrossRefGoogle Scholar
  2. Birzele, F., & Kramer, S. (2006). A new representation for protein secondary structure prediction based on frequent patterns. Bioinformatics, 22(21), 2628–2634. CrossRefGoogle Scholar
  3. Brain, D., & Webb, G. I. (2002). The need for low bias algorithms in classification learning from large data sets. In Proceedings of the sixth European conference on principles of data mining and knowledge discovery (PKDD) (pp. 62–73). Berlin: Springer. CrossRefGoogle Scholar
  4. Breiman, L. (2001). Random forests. Machine Learning, 45, 5–32. zbMATHCrossRefGoogle Scholar
  5. Camporelli, M. (2006). Using a Bayesian classifier for probability estimation: analysis of the AMIS score for risk stratification in myocardial infarction. Diploma thesis, Department of Informatics, University of Zurich. Google Scholar
  6. Cerquides, J., & Mántaras, R. L. D. (2005). Robust Bayesian linear classifier ensembles. In Proceedings of the sixteenth European conference on machine learning (pp. 70–81). Google Scholar
  7. Cestnik, B. (1990). Estimating probabilities: a crucial task in machine learning. In Proceedings of the ninth European conference on artificial intelligence (pp. 147–149). London: Pitman. Google Scholar
  8. Domingos, P., & Pazzani, M. J. (1996). Beyond independence: conditions for the optimality of the simple Bayesian classifier. In Proceedings of the thirteenth international conference on machine learning (pp. 105–112). San Mateo: Morgan Kaufmann. Google Scholar
  9. Fayyad, U. M., & Irani, K. B. (1993). Multi-interval discretization of continuous-valued attributes for classification learning. In Proceedings of the thirteenth international joint conference on artificial intelligence (pp. 1022–1029). San Mateo: Morgan Kaufmann. Google Scholar
  10. Ferrari, L. D., & Aitken, S. (2006). Mining housekeeping genes with a naive Bayes classifier. BMC Genomics, 7(1), 277. CrossRefGoogle Scholar
  11. Flikka, K., Martens, L., Vandekerckhove, J., Gevaert, K., & Eidhammer, I. (2006). Improving the reliability and throughput of mass spectrometry-based proteomics by spectrum quality filtering. Proteomics, 6(7), 2086–2094. CrossRefGoogle Scholar
  12. Flores, M., Gámez, J., Martínez, A., & Puerta, J. (2009). GAODE and HAODE: two proposals based on AODE to deal with continuous variables. In Proceedings of the 26th annual international conference on machine learning (pp. 313–320). New York: ACM. Google Scholar
  13. Friedman, N., Geiger, D., & Goldszmidt, M. (1997). Bayesian network classifiers. Machine Learning, 29(2), 131–163. zbMATHCrossRefGoogle Scholar
  14. Garcia, B., Aler, R., Ledezma, A., & Sanchis, A. (2008). Protein-protein functional association prediction using genetic programming. In Proceedings of the tenth annual conference on genetic and evolutionary computation (pp. 347–348). New York: ACM. CrossRefGoogle Scholar
  15. García-Jiménez, B., Juan, D., Ezkurdia, I., Andrés-León, E., & Valencia, A. (2010). Inference of functional relations in predicted protein networks with a machine learning approach. PLoS ONE, 4, e9969. CrossRefGoogle Scholar
  16. Hopfgartner, F., Urruty, T., Lopez, P., Villa, R., & Jose, J. (2010). Simulated evaluation of faceted browsing based on feature selection. Multimedia Tools and Applications, 47(3), 631–662. CrossRefGoogle Scholar
  17. Hunt, K. (2006). Evaluation of novel algorithms to optimize risk stratification scores in myocardial infarction. PhD thesis, Department of Informatics, University of Zurich. Google Scholar
  18. Jiang, L., & Zhang, H. (2006). Weightily averaged one-dependence estimators. In PRICAI 2006: trends in artificial intelligence (pp. 970–974). CrossRefGoogle Scholar
  19. Kunchevaa, L. I., Vilas, V. J. D. R., & Rodríguezc, J. J. (2007). Diagnosing scrapie in sheep: a classification experiment. Computers in Biology and Medicine, 37(8), 1194–1202. CrossRefGoogle Scholar
  20. Kurz, D., Bernstein, A., Hunt, K., Radovanovic, D., Erne, P., Siudak, Z., & Bertel, O. (2009). Simple point-of-care risk stratification in acute coronary syndromes: the AMIS model. British Medical Journal, 95(8), 662. Google Scholar
  21. Langley, P., & Sage, S. (1994). Induction of selective Bayesian classifiers. In Proceedings of the tenth conference on uncertainty in artificial intelligence (pp. 399–406). San Mateo: Morgan Kaufmann. Google Scholar
  22. Lasko, T. A., Atlas, S. J., Barry, M. J., & Chueh, K. H. C. (2006). Automated identification of a physician’s primary patients. Journal of the American Medical Informatics Association, 13(1), 74–79. CrossRefGoogle Scholar
  23. Lau, Q. P., Hsu, W., Lee, M. L., Mao, Y., & Chen, L. (2007). Prediction of cerebral aneurysm rupture. In Proceedings of the nineteenth IEEE international conference on tools with artificial intelligence (pp. 350–357). Washington: IEEE Computer Society. CrossRefGoogle Scholar
  24. Leon, A., et al. (2009). EcID. A database for the inference of functional interactions in E. coli. Nucleic Acids Research, 37, D629 (Database issue). CrossRefGoogle Scholar
  25. Liew, C., Ma, X., & Yap, C. (2010). Consensus model for identification of novel PI3K inhibitors in large chemical library. Journal of Computer-Aided Molecular Design, 24(2), 131–141. CrossRefGoogle Scholar
  26. Masegosa, A., Joho, H., & Jose, J. (2007). Evaluating query-independent object features for relevancy prediction. In Advances in information retrieval (pp. 283–294). CrossRefGoogle Scholar
  27. Mitchell, T. M. (1982). Generalization as search. Artificial Intelligence, 18(2), 203–226. MathSciNetCrossRefGoogle Scholar
  28. Nikora, A. P. (2005). Classifying requirements: towards a more rigorous analysis of natural-language specifications. In Proceedings of the sixteenth IEEE international symposium on software reliability engineering (pp. 291–300). Washington: IEEE Computer Society. Google Scholar
  29. Orhan, Z., & Altan, Z. (2006). Impact of feature selection for corpus-based WSD in Turkish. In Proceedings of the fifth Mexican international conference on artificial intelligence (pp. 868–878). Berlin: Springer. Google Scholar
  30. Pazzani, M. J. (1996). Constructive induction of Cartesian product attributes. In ISIS: information, statistics and induction in science (pp. 66–77). Google Scholar
  31. Sahami, M. (1996). Learning limited dependence Bayesian classifiers. In Proceedings of the second international conference on knowledge discovery in databases (pp. 334–338). Menlo Park: AAAI Press. Google Scholar
  32. Shahri, S., & Jamil, H. (2009). An extendable meta-learning algorithm for ontology mapping. In Flexible query answering systems (pp. 418–430). CrossRefGoogle Scholar
  33. Simpson, M., Demner-Fushman, D., Sneiderman, C., Antani, S., & Thoma, G. (2009). Using non-lexical features to identify effective indexing terms for biomedical illustrations. In Proceedings of the 12th conference of the European chapter of the association for computational linguistics (pp. 737–744). Association for Computational Linguistics. Google Scholar
  34. Tian, Y., Chen, C., & Zhang, C. (2008). Aode for source code metrics for improved software maintainability. In Fourth international conference on semantics, knowledge and grid (pp. 330–335). CrossRefGoogle Scholar
  35. Ting, K. M., Wells, J. R., Tan, S. C., Teng, S. W., & Webb, G. I. (2011). Feature-subspace aggregating: ensembles for stable and unstable learners. Machine Learning, 82(3), 375–397. CrossRefGoogle Scholar
  36. Wang, H., Klinginsmith, J., Dong, X., Lee, A., Guha, R., Wu, Y., Crippen, G., & Wild, D. (2007). Chemical data mining of the NCI human tumor cell line database. Journal of Chemical Information and Modeling, 47(6), 2063–2076. CrossRefGoogle Scholar
  37. Webb, G. I. (2000). Multiboosting: a technique for combining boosting and wagging. Machine Learning, 40(2), 159–196. CrossRefGoogle Scholar
  38. Webb, G. I., Boughton, J., & Wang, Z. (2005). Not so naive Bayes: aggregating one-dependence estimators. Machine Learning, 58(1), 5–24. zbMATHCrossRefGoogle Scholar
  39. Witten, I. H., & Frank, E. (2005). Data mining: practical machine learning tools and techniques. San Mateo: Morgan Kaufmann. zbMATHGoogle Scholar
  40. Yang, Y., Webb, G., Cerquides, J., Korb, K., Boughton, J., & Ting, K. M. (2006). To select or to weigh: a comparative study of model selection and model weighing for SPODE ensembles. In Proceedings of the seventeenth European conference on machine learning (pp. 533–544). Berlin: Springer. Google Scholar
  41. Yang, Y., Webb, G. I., Cerquides, J., Korb, K. B., Boughton, J., & Ting, K. M. (2007). To select or to weigh: a comparative study of linear combination schemes for superparent-one-dependence estimators. IEEE Transactions on Knowledge and Data Engineering, 19(12), 1652–1665. CrossRefGoogle Scholar
  42. Yang, Y., Webb, G. I., Korb, K., & Ting, K.-M. (2007). Classifying under computational resource constraints: anytime classification using probabilistic estimators. Machine Learning, 69(1), 35–53. CrossRefGoogle Scholar
  43. Zheng, Z., & Webb, G. I. (2000). Lazy learning of Bayesian rules. Machine Learning, 41(1), 53–84. CrossRefGoogle Scholar
  44. Zheng, F., & Webb, G. I. (2006). Efficient lazy elimination for averaged-one dependence estimators. In Proceedings of the twenty-third international conference on machine learning (pp. 1113–1120). New York: ACM. Google Scholar
  45. Zheng, F., & Webb, G. I. (2007). Finding the right family: parent and child selection for averaged one-dependence estimators. In Proceedings of the eighteenth European conference on machine learning (pp. 490–501). Berlin: Springer. Google Scholar

Copyright information

© The Author(s) 2011

Authors and Affiliations

  • Geoffrey I. Webb
    • 1
  • Janice R. Boughton
    • 1
  • Fei Zheng
    • 1
  • Kai Ming Ting
    • 1
  • Houssam Salem
    • 1
  1. 1.Faculty of Information TechnologyMonash UniversityClaytonAustralia

Personalised recommendations