Outlier Detection Ensemble with Embedded Feature Selection
Authors: Li Cheng, Yijie Wang, Xinwang Liu, Bin Li3503-3512
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Comprehensive experimental results on 12 real-world datasets from diverse domains validate the superiority of the proposed ODEFS. and Extensive empirical results on 12 real-world data sets show that ODEFS (i) reduces a large proportion of features and improves the performance of the original bare method; (ii) performs substantially better and more stably than the state-of-the-art competitors; (iii) has much better resilience to noisy features than its competitors; (iv) has linear time complexity w.r.t. data size and feature size. |
| Researcher Affiliation | Academia | Li Cheng,1 Yijie Wang,1* Xinwang Liu, Bin Li1 1Science and Technology on Parallel and Distributed Processing Laboratory College of Computer, National University of Defense Technology Changsha, China {chengli09, wangyijie, liuxinwang, libin16a}@nudt.edu.cn |
| Pseudocode | Yes | Algorithm 1 ODEFS |
| Open Source Code | No | No statement regarding the release of source code for ODEFS or a link to a code repository was found. |
| Open Datasets | Yes | They are available at http://archive.ics.uci.edu/ml/index.php, http://odds.cs.stonybrook.edu/, http://vision.cs.uiuc.edu/attributes/ |
| Dataset Splits | No | The paper mentions 'training set' but does not specify explicit train/validation/test splits (e.g., percentages, sample counts) for the datasets used in their experiments. |
| Hardware Specification | Yes | All the experiments are executed at a PC in a 3.6GHz CPU with 16GB memory. |
| Software Dependencies | Yes | ODEFS and its competitors are implemented in Python 3.4. |
| Experiment Setup | Yes | In our experiments, ODEFS uses m = 32 for small datasets (i.e., n ≤ 10^4) and m = 64 for large datasets (i.e., n > 10^4). Other parameters setting, i.e., a = 2, m' = 6m, l = 2*sqrt(n) has been explained in the above sections. The parameters of Le Si NN are set as the recommended settings. and epsilon = 0.05 is a small constant |