Dropout Training for Support Vector Machines
Authors: Ning Chen, Jun Zhu, Jianfei Chen, Bo Zhang
AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Empirical results on several real datasets demonstrate the effectiveness of dropout training on significantly boosting the classification accuracy of linear SVMs. |
| Researcher Affiliation | Academia | Ning Chen Jun Zhu Jianfei Chen Bo Zhang State Key Lab of Intelligent Tech. & Systems; Tsinghua National TNList Lab; Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China {ningchen@mail, dcszj@mail, chenjf10@mails, dcszb@mail}.tsinghua.edu.cn |
| Pseudocode | No | The paper describes the Iteratively Re-weighted Least Square Algorithm verbally but does not present it in a structured pseudocode block or algorithm box. |
| Open Source Code | No | The paper states, "We implement both Dropout-SVM and Dropout-Logistic using C++," but it does not provide any link or explicit statement about the availability of the source code for the described methodology. |
| Open Datasets | Yes | We use the public Amazon book review and kitchen review datasets (Blitzer, Dredze, and Pereira 2007)... We choose the CIFAR-10 image categorization dataset8. (http://www.cs.toronto.edu/ kriz/cifar.html)... We choose the the MNIST dataset, which consists of 60,000 training and 10,000 testing handwritten digital images from 10 categories |
| Dataset Splits | Yes | The hyper-parameters are selected via cross-validation on the training set. ... During training, we choose the best models over different dropout levels via cross-validation. |
| Hardware Specification | No | The paper states, "We implement both Dropout-SVM and Dropout-Logistic using C++..." but does not specify any particular hardware components such as GPU or CPU models, memory, or specific computing environments used for the experiments. |
| Software Dependencies | No | The paper mentions implementation "using C++" but does not provide specific version numbers for the programming language or any other ancillary software dependencies like libraries or frameworks. |
| Experiment Setup | Yes | We consider the unbiased dropout (or blankout) noise model6, that is, p( x = 0) = q and p( x = 1 1 qx) = 1 q, where q 2 [0, 1) is a pre-specified corruption level. ... for each value of M we choose the dropout model with q selected by cross-validation. The hyper-parameter of the SVM classifier is also chosen via cross-validation on the training data. ... The hyper-parameters are selected via cross-validation on the training set. |