Summary of A Neighbor-searching Discrepancy-based Drift Detection Scheme For Learning Evolving Data, by Feng Gu et al.
A Neighbor-Searching Discrepancy-based Drift Detection Scheme for Learning Evolving Data
by Feng Gu, Jie Lu, Zhen Fang, Kun Wang, Guangquan Zhang
First submitted to arxiv on: 23 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Machine learning models need to adapt quickly in response to changing data streams without sacrificing performance. One major challenge is real concept drift, which causes classification performance to deteriorate over time. Existing drift detection methods, such as two-sample distribution tests and monitoring classification error rates, have limitations like high computational cost or inability to distinguish between real and virtual drifts. Moreover, none of these methods provide information on the trend of the drift, which is crucial for model maintenance. This paper proposes a novel method called Neighbor-Searching Discrepancy, which measures the classification boundary difference between two samples. The approach detects real concept drift with high accuracy while ignoring virtual drift and indicates the direction of boundary change by identifying class invasion or retreat. This information can help determine separability changes between classes. The method is evaluated through 11 experiments on artificial and real-world datasets, showing robustness against various distributions and dimensions, outperforming state-of-the-art methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Machine learning models need to adapt quickly in response to changing data streams without sacrificing performance. One major challenge is real concept drift, which causes classification performance to deteriorate over time. This paper proposes a new way to detect this problem using an approach called Neighbor-Searching Discrepancy. This method measures the difference between two samples and can accurately detect changes in the data stream while ignoring irrelevant changes. It also provides information on the direction of the change, which is important for model maintenance. |
Keywords
» Artificial intelligence » Classification » Machine learning
