Open Category Classification by Adversarial Sample Generation
Authors: Yang Yu, Wei-Yang Qu, Nan Li, Zimin Guo
IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 4 Experiments. Experiments performed on several datasets show the effectiveness of ASG. |
| Researcher Affiliation | Collaboration | Yang Yu1, Wei-Yang Qu1, Nan Li2, Zimin Guo3 1National Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, China 1Collaborative Innovation Center of Novel Software Technology and Industrialization, Nanjing, China 2Alibaba Group, Hangzhou, China 3College of Engineering, UC Berkeley |
| Pseudocode | Yes | Algorithm 1 Generation of negative instances of seen classes |
| Open Source Code | No | The paper provides a link to an implementation of the RACOS algorithm (https://github.com/eyounx/ZOOpt), which is a derivative-free optimization method used in their approach, but does not provide concrete access to the source code for their proposed ASG framework. |
| Open Datasets | Yes | Handwritten Digit Image Classification: We conducted the second experiment on the MNIST handwritten digit dataset. Document Classification: In the third experiment, we conduct experiments on the 20 Newsgroups dataset, which is a popular text dataset consists of documents from 20 different topics. |
| Dataset Splits | Yes | The coefficient C in SVM is determined by cross validation on the training dataset using the original OVR-SVM. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU, GPU models, or cloud instance types) used for running the experiments. |
| Software Dependencies | No | The paper mentions 'LIBSVM software' and 'RACOS algorithm1 [Yu et al., 2016] with default parameters. 1using the implementation in https://github.com/eyounx/ZOOpt' but does not specify version numbers for these software dependencies (LIBSVM, ZOOpt). |
| Experiment Setup | Yes | The coefficient C in SVM is determined by cross validation on the training dataset using the original OVR-SVM. The width of Gaussian kernel γ is fixed to 1/d, where d is the size of the feature. For the ASG algorithm, the distance measure dist is set to be the Euclidean distance. When generating negative samples of class k, the parameter C1,C2 is set as minx1,x2 Dk;x1 =x2 dist(x1, x2), λ1 and λ2 are both set to 0.1, and T = 200. When generating positive samples for class k, the parameter C3 is also set to minx1,x2 Dk;x1 =x2 dist(x1, x2), η is set to 0.3, and also T = 200. |