Agnostic Bayesian Learning of Ensembles

Authors: Alexandre Lacoste, Mario Marchand, François Laviolette, Hugo Larochelle

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, several experimental results are presented in Section 6. We performed experiments to assess the performance of the agnostic Bayes ensemble approach and compared with a few commonly used methods:
Researcher Affiliation Academia Alexandre Lacoste D epartement d informatique et de g enie logiciel, Universit e Laval, Qu ebec, Canada, G1K-7P4 Hugo Larochelle D epartement d informatique, Universit e de Sherbrooke, Qu ebec, Canada, J1K-2R1 Mario Marchand D epartement d informatique et de g enie logiciel, Universit e Laval, Qu ebec, Canada, G1K-7P4 Franc ois Laviolette D epartement d informatique et de g enie logiciel, Universit e Laval, Qu ebec, Canada, G1K-7P4
Pseudocode No The paper describes algorithmic steps in prose but does not provide any formally structured pseudocode blocks or algorithms labeled as such.
Open Source Code No The paper does not provide any explicit statement or link for open-source code availability for the methodology described.
Open Datasets Yes To build a substantial collection of datasets, we used the AYSU collection (Ulas et al., 2009) coming from the UCI and the Delve repositories and we added the MNIST dataset. We have also collected 22 regression datasets from the Louis Torgo collection.3 These datasets were obtained from the following source : http://www.dcc.fc.up.pt/ ltorgo/Regression/ Data Sets.html
Dataset Splits Yes A common approach to this problem is to estimate the generalization performance of each predictor on a holdout dataset (through a training/validation set split or using kfold cross-validation) and use the predictor with the best performance. Let {V1, V2, . . . , Vk} be a partition of S, and let hγ,j def= Aγ (S \ Vj). Traditional cross validation is used to select the best soft margin parameter
Hardware Specification No The paper mentions 'Calcul Qu ebec for providing support and access to Colosse s high performance computer grid' but does not specify exact hardware components like GPU/CPU models or memory details.
Software Dependencies No Except for a custom implementation of ANN and KRR, we used scikit-learn (Pedregosa et al., 2011) for all other implementations.
Experiment Setup Yes The set Γ of models used in this experiment is a combination of SVMs, Artificial Neural Networks (ANN), random forests, extra randomized trees (Geurts et al., 2006) and gradient tree boosting (Friedman, 2001) with several variants of hyperparameters. Traditional cross validation is used to select the best soft margin parameter over 20 candidates values ranging from 10 3 to 100 on a logarithmic scale. We use 1000 samples from p(r|L) to estimate p(h|S).