Angle-based Multicategory Distance-weighted SVM
Authors: Hui Sun, Bruce A. Craig, Lingsong Zhang
JMLR 2017 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Theoretical and numerical studies demonstrate the advantages of MDWSVM method over existing angle-based methods. ... In Section 4, we perform simulation studies to compare our model with MSVM and MDWD. ... Section 5 involves a real application. |
| Researcher Affiliation | Academia | Hui Sun EMAIL Department of Statistics Purdue University West Lafayette, IN 47906, USA Bruce A. Craig EMAIL Department of Statistics Purdue University West Lafayette, IN 47906, USA Lingsong Zhang EMAIL Department of Statistics Purdue University West Lafayette, IN 47906, USA |
| Pseudocode | No | The paper describes mathematical formulations and optimization problems, but it does not contain any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | Model (6) can be easily implemented in Matlab using the CVX package (Grant et al., 2008). ... A similar implementation for our MDWSVM method will be explored in the future. |
| Open Datasets | Yes | In this section, we apply our MDWSVM method to a real data used in Shen and Huang (2005). |
| Dataset Splits | Yes | In each example, we simulate a training data set, a tuning data set, and a testing data set. The training data and tuning data have the same sample sizes and are used to estimate the model and to find the optimal tuning parameters. The size of the testing data set is ten times the size of the training data... The data are divided into three groups, Group 1: Monday (size 48); Group 2: Tuesday to Thursday (size 151); Group 3: Friday (size 52). This dataset is a typical imbalanced HDLSS dataset with Group 2 as the dominant group. ... all three methods use five-fold cross validation. |
| Hardware Specification | Yes | All numerical experiments were carried out on an Intel Xeon E3-1284L (2.5 GHz) processor. |
| Software Dependencies | No | Model (6) can be easily implemented in Matlab using the CVX package (Grant et al., 2008). While MATLAB and CVX are mentioned, specific version numbers for these software components are not provided. |
| Experiment Setup | Yes | There are two parameters C and α in our MDWSVM method. ... We set α to be fixed, varied C... We let α = 0.5, and allow C to change from 2^-3 to 2^12... One can simply fix α = 0.5 and use cross-validation to choose C. This is why we fix α = 0.5 in our simulation. |