An evaluation method and device of a polytomy algorithm, an electronic device, and a storage medium
By automatically identifying positive samples in the clustering algorithm and using metrics such as accuracy, recall, and F1 score, the problem of high manual cost and single metric in the evaluation of the clustering algorithm is solved, and qualitative comparison and evaluation on unlabeled datasets are realized.
CN120726427BActive Publication Date: 2026-07-24ZHEJIANG UNIVIEW TECH CO LTD
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHEJIANG UNIVIEW TECH CO LTD
- Filing Date
- 2024-03-29
- Publication Date
- 2026-07-24
AI Technical Summary
Technical Problem
In existing evaluation methods for clustering algorithms, using labeled datasets is costly and inefficient, while using unlabeled datasets results in limited metric dimensions and a lack of interpretability.
Method used
By determining the clustering results of the first and second clustering algorithms, the system automatically identifies positive samples of the target image in different sets and uses metrics such as accuracy, recall, and F1 score to perform a qualitative comparison of the clustering effect.
Benefits of technology
This study enables qualitative comparison of clustering algorithms on unlabeled datasets, saving the cost of manual annotation and accurately evaluating clustering performance through quantitative indicators.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN120726427B_ABST
Abstract
The application discloses a kind of evaluation method, device, electronic equipment and storage medium of poly-filing algorithm.It includes: determining the first poly-filing result obtained by first poly-filing algorithm to target data set and the second poly-filing result obtained by second poly-filing algorithm to target data set;Determine the target image in target data set, and determine the first set of target image in the first poly-filing result and the second set of target image in the second poly-filing result;According to first poly-filing algorithm, second poly-filing algorithm, first set and second set, respectively determine the positive sample in first set and the positive sample in second set;According to the positive sample in the first set corresponding to at least two target images, determine the evaluation value of first poly-filing algorithm, according to the positive sample in the second set corresponding to at least two target images, determine the evaluation value of second poly-filing algorithm.The application can qualitatively compare the effect of poly-filing algorithm, save artificial cost.
Need to check novelty before this filing date? Find Prior Art