Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Neighborhood Consensus Networks for Unsupervised Multi-view Outlier Detection
Authors: Li Cheng, Yijie Wang, Xinwang Liu7099-7106
AAAI 2021 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results show that our method significantly outperforms state-of-the-art methods. |
| Researcher Affiliation | Academia | Li Cheng, Yijie Wang , Xinwang Liu College of Computer, National University of Defense Technology, Changsha, China EMAIL |
| Pseudocode | Yes | Algorithm 1 Optimization of NCMOD |
| Open Source Code | Yes | All the source codes, including the proposed NCMOD, the competitors and datasets generation, are provided in the supplementary. |
| Open Datasets | Yes | Three widely used high dimensional benchmarks are used: MNIST, REUTERS, TTC, and the dimensionality of them are respectively 784, 2000 and 7507. |
| Dataset Splits | No | The paper describes how the datasets are generated and structured for outlier detection but does not specify a separate validation dataset split in terms of percentages or counts for hyperparameter tuning or model selection in the traditional sense of supervised learning. It mentions setting a threshold for F1-Score evaluation, but not a validation split. |
| Hardware Specification | Yes | The methods are all implemented in Python 3.4 executed at a PC in a 3.6GHz CPU with 16GB memory. |
| Software Dependencies | No | The paper states 'implemented in Python 3.4'. While Python 3.4 is a specific software version, it does not list any other key software components, libraries, or solvers with their version numbers, which would typically be needed for full reproducibility of ancillary software. |
| Experiment Setup | Yes | We set k = 8 and α = 1.0 as default. The threshold is set as the outlier ratio, i.e., 0.15. |