Hierarchical classification with multi-path selection based on granular computing
- PDF / 2,205,673 Bytes
- 23 Pages / 439.37 x 666.142 pts Page_size
- 44 Downloads / 219 Views
Hierarchical classification with multi‑path selection based on granular computing Shunxin Guo1 · Hong Zhao1,2
© Springer Nature B.V. 2020
Abstract Hierarchical classification is a research hotspot in machine learning due to the widespread existence of data with hierarchical class structures. Existing hierarchical classification methods based on granular computing can effectively reduce the computational complexity by considering the granularity of classes. However, their predictive accuracy is affected by inter-level error propagation within the hierarchy. In this paper, we propose a hierarchical classification method with multi-path selection based on coarse- and fine-grained class relationships, which mitigates the inter-level error propagation problem. Firstly, we use a top-down recursive method to calculate the probabilities of the hierarchical classes by logistic regression classification. Secondly, the current class probability is calculated by combining the parent and current classes probabilities. We select multiple possible finegrained classes at the current level according to their sibling relationships. Compared with existing methods, the proposed method reduces the possibility of misclassification from the upper layer. Finally, the multi-path prediction result is provided to a classical classifier for final prediction. Our hierarchical classification method is evaluated on six benchmark datasets to demonstrate that it provides better classification performance than existing state-ofthe-art hierarchical methods. Keywords Granular computing · Hierarchical classification · Inter-level error propagation · Multi-path selection
1 Introduction Granular computing is a powerful tool for analyzing, understanding, representing, and solving real-world problems at different granularity levels (Tan et al. 2020; Yao 2013; Dai et al. 2018; Zhang et al. 2020). The granular computing approach can effectively solve * Hong Zhao [email protected] Shunxin Guo [email protected] 1
Fujian Key Laboratory of Granular Computing and Application, Minnan Normal University, Zhangzhou 363000, Fujian, China
2
School of Computer Science, Minnan Normal University, Zhangzhou 363000, Fujian, China
13
Vol.:(0123456789)
S. Guo, H. Zhao
large-scale classification tasks according to levels of semantic granularity. In this approach, data have different granularities, from coarse-grained to fine-grained. Large-scale classification tasks can be divided into a series of small-scale and easy-to-handle sub-tasks. Hierarchical classification (Alshamaa et al. 2018) has received increasing attention in the fields of data mining and machine learning. Recently, many hierarchical classification methods have been proposed for different applications, such as the classification of text, image, audio, and biological data. Ghazi et al. (2010) constructed a class hierarchy based on degrees of emotion for automatic text classification. This hierarchical approach can alleviate the effect of imbalanced data on text classification. For larges
Data Loading...
