Improved Subsampled Randomized Hadamard Transform for Linear SVM
Authors: Zijian Lei, Liang Lan4519-4526
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our experimental results have demonstrated that our proposed methods can achieve higher classification accuracies than SRHT and other random projection methods on six real-life datasets. Finally, we performed experiments to evaluate our proposed methods on six real-life datasets. Our experimental results clearly demonstrate that our proposed method can obtain higher classification accuracies than SRHT and other three popular random projection methods while only slightly increasing the running time. The accuracy and training time of all algorithms are reported in Table 2. |
| Researcher Affiliation | Academia | Zijian Lei Department of Computer Science, Hong Kong Baptist University, Hong Kong SAR, China cszjlei@comp.hkbu.edu.hk Liang Lan Department of Computer Science, Hong Kong Baptist University, Hong Kong SAR, China lanliang@comp.hkbu.edu.hk |
| Pseudocode | Yes | Algorithm 1 Improved Subsampled Randomized Hadamard Transform (ISRHT) for linear SVM classification |
| Open Source Code | No | The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available. |
| Open Datasets | Yes | These benchmark datasets are downloaded from LIBSVM website 2. https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/ |
| Dataset Splits | Yes | For datasets (mushrooms,real-sim, rcv1-binary, news20-binary), we refer to the setting in (Pourkamali-Anaraki, Becker, and Wakin 2018) and randomly select 70% of the original training data as training data and the rest 30% as test data. The regularization parameter C in linear SVM is chosen from {2^-5, 2^-4, ..., 2^4, 2^5} by 5-fold cross validation. |
| Hardware Specification | Yes | Our experiments are performed on a server with Dual 6-core Intel Xeon 2.4GHz CPU and 128 GB RAM. |
| Software Dependencies | No | The paper mentions using "liblinear" but does not specify its version number or any other software dependencies with version information. |
| Experiment Setup | Yes | The regularization parameter C in linear SVM is chosen from {2^-5, 2^-4, ..., 2^4, 2^5} by 5-fold cross validation. The tradeoff parameter a for ISRHT-supervised is fixed to 1.0. |