# On PAC-Bayesian bounds for random forests

- 460 Downloads

**Part of the following topical collections:**

## Abstract

Existing guarantees in terms of rigorous upper bounds on the generalization error for the original random forest algorithm, one of the most frequently used machine learning methods, are unsatisfying. We discuss and evaluate various PAC-Bayesian approaches to derive such bounds. The bounds do not require additional hold-out data, because the out-of-bag samples from the bagging in the training process can be exploited. A random forest predicts by taking a majority vote of an ensemble of decision trees. The first approach is to bound the error of the vote by twice the error of the corresponding Gibbs classifier (classifying with a single member of the ensemble selected at random). However, this approach does not take into account the effect of averaging out of errors of individual classifiers when taking the majority vote. This effect provides a significant boost in performance when the errors are independent or negatively correlated, but when the correlations are strong the advantage from taking the majority vote is small. The second approach based on PAC-Bayesian \(C\)-bounds takes dependencies between ensemble members into account, but it requires estimating correlations between the errors of the individual classifiers. When the correlations are high or the estimation is poor, the bounds degrade. In our experiments, we compute generalization bounds for random forests on various benchmark data sets. Because the individual decision trees already perform well, their predictions are highly correlated and the \(C\)-bounds do not lead to satisfactory results. For the same reason, the bounds based on the analysis of Gibbs classifiers are typically superior and often reasonably tight. Bounds based on a validation set coming at the cost of a smaller training set gave better performance guarantees, but worse performance in most experiments.

## Keywords

PAC-Bayesian analysis Random forests Majority vote## Notes

### Acknowledgements

We acknowledge support by the Innovation Fund Denmark through the *Danish Center for Big Data Analytics Driven Innovation* (DABAI).

## Supplementary material

## References

- Andersen, M., & Vandenberghe, L. (2019). CVXOPT. Retrieved May, 2019 from http://cvxopt.org/.
- Arlot, S., & Genuer, R. (2014). Analysis of purely random forests bias. arXiv:1407.3939.
- Biau, G. (2012). Analysis of a random forests model.
*Journal of Machine Learning Research*,*13*(1), 1063–1095.MathSciNetzbMATHGoogle Scholar - Biau, G., Devroye, L., & Lugosi, G. (2008). Consistency of random forests and other averaging classifiers.
*Journal of Machine Learning Research*,*9*, 2015–2033.MathSciNetzbMATHGoogle Scholar - Breiman, L. (1996b). Out-of-bag estimation. Retrieved May, 2019 from https://www.stat.berkeley.edu/users/breiman/OOBestimation.pdf.
- Breiman, L. (2002). Some infinity theory for predictor ensembles.
*Journal of Combinatorial Theory A*,*98*, 175–191.CrossRefGoogle Scholar - Denil, M., Matheson, D., & Freitas, N. D. (2014). Narrowing the gap: Random forests in theory and in practice. In
*Proceedings of the 31st international conference on machine learning (ICML), PMLR, Proceedings of machine learning research*(Vol. 32, pp. 665–673).Google Scholar - Fernández-Delgado, M., Cernadas, E., Barro, S., & Amorim, D. (2014). Do we need hundreds of classifiers to solve real world classification problems?
*Journal of Machine Learning Research*,*15*, 3133–3181.MathSciNetzbMATHGoogle Scholar - Genuer, R. (2010). Risk bounds for purely uniformly random forests. Technical report. France: Institut National de Recherche en Informatique et en Automatique.Google Scholar
- Germain, P., Lacasse, A., Laviolette, F., Marchand, M., & Roy, J. F. (2015). Risk bounds for the majority vote: From a PAC-Bayesian analysis to a learning algorithm.
*Journal of Machine Learning Research*,*16*, 787–860.MathSciNetzbMATHGoogle Scholar - Geurts, P., Ernst, D., & Wehenkel, L. (2006). Extremely randomized trees.
*Machine Learning*,*63*(1), 3–42.CrossRefzbMATHGoogle Scholar - Gieseke, F., & Igel, C. (2018). Training big random forests with little resources. In
*Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery and data mining (KDD)*(pp. 1445–1454). ACM Press.Google Scholar - Hastie, T., Tibshirani, R., & Friedman, J. (2009).
*The elements of statistical learning*(2nd ed.). Berlin: Springer.CrossRefzbMATHGoogle Scholar - Langford, J., & Shawe-Taylor, J. (2002). PAC-Bayes & Margins. In
*Proceedings of the 15th international conference on neural information processing systems*(pp. 439–446). MIT Press.Google Scholar - Maurer, A. (2004). A note on the PAC-Bayesian theorem. arXiv:cs/0411099.
- Mcallester, D. (2003). Simplified PAC-Bayesian margin bounds. In
*Proceedings of the 16th annual conference on computational learning theory (COLT). LNCS*(Vol. 2777, pp. 203–215). Springer.Google Scholar - McAllester, D. A. (1998). Some PAC-Bayesian theorems. In
*Proceedings of the eleventh annual conference on computational learning theory (COLT)*(pp. 230–234). ACM.Google Scholar - McAllester, D. A. (1999). PAC-Bayesian model averaging. In
*Proceedings of the twelfth annual conference on computational learning theory (COLT)*(pp. 164–170). ACM.Google Scholar - Oneto, L., Cipollini, F., Ridella, S., & Anguita, D. (2018). Randomized learning: Generalization performance of old and new theoretically grounded algorithms.
*Neurocomputing*,*298*, 21–33.CrossRefGoogle Scholar - Schapire, R. E., & Singer, Y. (1999). Improved boosting algorithms using confidence-rated predictions.
*Machine Learning*,*37*(3), 297–336.CrossRefzbMATHGoogle Scholar - Seeger, M. (2002). PAC-Bayesian generalization error bounds for Gaussian process classification.
*Journal of Machine Learning Research*,*3*, 233–269.CrossRefzbMATHGoogle Scholar - Thiemann, N., Igel, C., Wintenberger, O., & Seldin, Y. (2017). A strongly quasiconvex PAC-Bayesian bound. In
*Proceedings of the international conference on algorithmic learning theory (ALT), PMLR, Proceedings of machine learning research*(Vol. 76, pp. 466–492).Google Scholar - Valiant, L. G. (1984). A theory of the learnable.
*Communications of the ACM*,*27*(11), 1134–1142.CrossRefzbMATHGoogle Scholar - Wang, Y., Tang, Q., Xia, S.T., Wu, J., & Zhu, X. (2016). Bernoulli random forests: Closing the gap between theoretical consistency and empirical soundness. In
*Proceedings of the 17th international conference on machine learning (ICML)*(pp. 2167–2173). Morgan Kaufmann.Google Scholar