Supervised Learning via Euler's Elastica Models

Authors: Tong Lin, Hanlin Xue, Ling Wang, Bo Huang, Hongbin Zha

JMLR 2015 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments have demonstrated the effectiveness of the proposed model for binary classification, multi-class classification, and regression tasks.
Researcher Affiliation Academia Key Laboratory of Machine Perception (Ministry of Education) School of Electronics Engineering and Computer Science Peking University, Beijing, 100871, China
Pseudocode No The paper describes numerical algorithms in text within Section 4 ('Numerical Algorithms'), but does not provide structured pseudocode blocks or algorithms.
Open Source Code No The paper mentions using 'LIBSVM implementation (Chang and Lin, 2011)' and 'Matlab neural network toolbox', which are third-party tools. It does not provide explicit statements or links for the authors' own implementation code.
Open Datasets Yes We collected real data sets from the libsvm website (Chang and Lin, 2011) and the UCI machine learning repository (Asuncion and Newman, 2013).
Dataset Splits Yes The optimal parameters for each algorithm are selected by grid search using 5-fold cross-validation. ... For each data set, we randomly run the 5-fold cross validation ten times to reduce the influence of data partitions.
Hardware Specification Yes The experiments are conducted on a PC Sever with two Intel Xeon 5620 cores and 8GB RAM.
Software Dependencies No The paper mentions 'LIBSVM implementation (Chang and Lin, 2011)' and 'Back-Propagation Neural Networks (BPNN) in the Matlab neural network toolbox' but does not specify version numbers for these software components.
Experiment Setup Yes The optimal parameters for each algorithm are selected by grid search using 5-fold cross-validation. ... only two common parameters are searched for all methods except BPNN: (C, g) for SVM, while (c, λ) for LR, TV, and EE. Empirically, the parameter η is set as 1 for LR, and the parameter b is fixed as 0.01 for EE. ... the two common parameters are searched from 10 : 10 in logarithm with step 2. The maximum number of iterations in GD and LAG is empirically setting as 40. All data sets are scaled into [0,1] before training and testing.