Discriminative Vanishing Component Analysis

Authors: Chenping Hou, Feiping Nie, Dacheng Tao

AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results are provided for demonstration. We would like to provide two groups of experimental results for illustration.
Researcher Affiliation Academia College of Science, National University of Defense Technology; Center for OPTical IMagery Analysis and Learning (OPTIMAL), Northwestern Polytechnical University; Center for Quantum Computation and Intelligent Systems and the Faculty of Engineering and Information Technology, University of Technology, Sydney
Pseudocode Yes Algorithm 1 Discriminative Vanishing Component Analysis (DVCA)
Open Source Code No The paper mentions that 'VCA is implemented by the code provided by the authors1. 1http://www.cs.huji.ac.il/ rlivni73/', which refers to the code for the baseline VCA, not the authors' own DVCA method.
Open Datasets Yes All the data are downloaded from open sources2,3. 2http://www.csie.ntu.edu.tw/ cjlin/libsvmtools/datasets/ 3http://www.cad.zju.edu.cn/home/dengcai/Data/data.html
Dataset Splits Yes The original data are randomly split into two parts, training and testing samples. The common parameter d is tuned using 5-fold cross validation.
Hardware Specification Yes We have tested the algorithm by a naive Matlab implementation on a workstation with 12 processor (3.33G for each) and 47.2GB memory.
Software Dependencies No The paper mentions 'Matlab implementation' and 'Lib Svm software (Chang and Lin 2011)' but does not provide specific version numbers for these or any other ancillary software components.
Experiment Setup No The paper states that 'The common parameter d is tuned using 5-fold cross validation. Besides, in DVCA, we also determine λ by cross validation.', but does not provide specific values for hyperparameters or detailed training configurations for the models used.