Anomaly Detection by Leveraging Incomplete Anomalous Knowledge with Anomaly-Aware Bidirectional GANs
Authors: Bowen Tian, Qinliang Su, Jian Yin
IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experimental results demonstrate that our proposed method is able to effectively make use of the incomplete anomalous information, leading to significant performance gains compared to existing methods. |
| Researcher Affiliation | Academia | 1School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China 2Guangdong Key Laboratory of Big Data Analysis and Processing, Guangzhou, China 3School of Artificial Intelligence, Sun Yat-sen University, Guangdong, China tianbw@mail2.sysu.edu.cn, {suqliang,issjyin}@mail.sysu.edu.cn |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Code is available at https://github.com/tbw162/AA-BiGAN. |
| Open Datasets | Yes | Following the paper [Ruff et al., 2019], for each dataset, we select one category as normal, while treating the remaining nine types as anomalies. The proposed model is trained on the normal samples and collected anomalies. ...on MNIST, F-MNIST and CIFAR-10... on six other classic anomaly detection datasets. Table 2 shows the performance of our proposed model under the scenario of γl = 0.01 and γp = 0. From the table, it can be seen our proposed model overall outperforms current baseline methods. Even on Arrhythimia dataset, which contains less than 500 samples, our model still achieve a 2% performance improvement, demonstrating the competitiveness of the proposed method on small datasets. |
| Dataset Splits | Yes | The testing dataset is splitted into a validation and testing dataset with a ratio of 20% and 80%. The hyperparameters are finetuned on the validation dataset, please refer to the Supplementary Material for more details of training. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details (e.g., library or solver names with version numbers) needed to replicate the experiment. |
| Experiment Setup | No | The paper states that hyperparameters are finetuned and refers to the Supplementary Material for more details of training, but does not provide these specific details within the main text. |