A New Generalized Error Path Algorithm for Model Selection

Authors: Bin Gu, Charles Ling

ICML 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experimental results on a variety of datasets not only confirm our theoretical findings, but also show that the best model with our GEP has better generalization error on the test data, compared to the grid search, manual search, and random search.
Researcher Affiliation Academia Bin Gu1,2 JSGUBIN@NUIST.EDU.CN Charles X. Ling1 CLING@CSD.UWO.CA 1Department of Computer Science, University of Western Ontario, Canada 2School of Computer & Software, Nanjing University of Information Science & Technology, China
Pseudocode Yes Algorithm 1 GEP (Generalized error path algorithm) ... Algorithm 2 CV-GEP (Cross validation with GEP)
Open Source Code No The paper does not provide concrete access to source code for the methodology described in this paper. It mentions using a third-party implementation for C-SVC: 'For the solution path of C-SVC, we used the implementation in http://web.eecs.umich.edu/ cscott/code.html#svmpath'.
Open Datasets Yes The Ionosphere, Diabetes, Hill-Valley, Breast Cancer, Housing, Forest Fires, Auto MPG, and Triazines datasets are from the UCI benchmark repository (Bache & Lichman, 2013).
Dataset Splits Yes We randomly partition each dataset into 65% training and 35% test sets. For each dataset, the training set is used with a 5-fold CV procedure to determine the optimal parameter.
Hardware Specification No The paper does not provide specific hardware details used for running its experiments.
Software Dependencies No The paper mentions 'We implemented our GEP in MATLAB' but does not provide specific version numbers for MATLAB or any other software dependencies.
Experiment Setup Yes GS is done on a τ grid linearly spaced in the region {log2 λ| 20 log2 λ 20}. ... the Gaussian kernel K(x1, x2) = exp( κ x1 x2 2) with κ = 0.5 was used.