Three-way decisions based blocking reduction models in hierarchical classification
Introduction
Hierarchical classification (HC) is an effective method to solve multiclass classification problems, especially when categories are organized hierarchically. Many important real-world classification problems naturally can be treated as HC problems, such as text categorization [4], [14], protein function prediction [27], [33], image classification [2], [6] etc. Using HC methods, a large-scale classification task can be divided into several small-scale tasks, so as to reduce the difficulty of classification. Usually, HC method uses a top-down level-based strategy, in which the class hierarchy typically stored as a tree. Such a strategy is simple, intuitive, and interpretable. However, due to the complexity of category relationships, samples are easily misclassified in higher-level classifiers, i.e. blocking. It is clear that blocking is one of reasons that HC performs worse than the traditional flat classification.
Most existing blocking reduction strategies in HC can be divided into two types depending on whether changing the category hierarchy: methods that will change the hierarchy and methods that will not change the hierarchy. For the first type, there are Restricted Voting method (RVM) [37], Priori Knowledge Based Hierarchical Classification method (PKHC) [40], data-driven hierarchical structure modification approach (Global-INF) [25], and etc. These strategies give the blocked samples another chance to return to the correct categories by changing the original hierarchy. For the second type, there are Multiplicative method (MM) [7], Extended Multiplicative method (EMM) [37], Threshold Reduction Method (TRM) [37], and etc. These strategies work on probabilities of base classifiers, taking into account the horizontal or vertical base classifiers’ results and helping blocked samples return to the correct category. In fact, methods of the second type have little improvement in the blocking problem. Since blocking is mainly caused by the complexity of category relationships, methods which change the hierarchy are more suitable for the blocking reduction problems. The two blocking reduction strategies proposed in this paper also belong to the first type.
Although there are many methods of the first type have been presented in the past, less of them considers the inconsistency between the artificially defined hierarchy and the actual hierarchy of the data. Some of existing methods take into account the relationship between base classifiers, but overrefine the hierarchical topological structure, which bring new blocking. Take PKHC method as an example, let us consider the most extreme case, high-level category H1 has two sub-categories L11 and L12, and all samples of L11 are misclassified as H2 at the high-level of the hierarchy, while all samples of L12 are classified correcetly as H1, see Fig. 2. PKHC thinks that samples of H1 are easily to be misclassified as H2, thus it will add both paths from H2 to L11 and L12, it increases training costs and increases the uncertainty of classifier . Actually, only the path from H2 to L11 need to be added. See Section 3 for more details.
To this end, in this study, we propose a new method for blocking reduction problems, the illustration of this method see Fig. 1. In a nutshell, the core of our method is to reconstruct the topology of the hierarchical classification model by learning category relations from the original category hierarchy, then using 3WD method to divide observed objects into three regions: positive region, boundary region and negative region, specially paying attention to the boundary region. We utilize two different methods to mine category relations. The first method takes into account the cross-level blocking priori knowledge based on the method proposed in paper [40], which only considers the relationships between high-level categories but neglects the relationships between high- and low-level categories. However, this method cares only the one-to-one relationships between categories, the second method applies the topic model based label grouping algorithm proposed in paper [39] to learn the many-to-many relationships amongst categories.
The proposed TriHC method can effectively reduce the blocking error and has less impact on categories that are not prone to blocking. Contributions of this study could be summarized as follows.
- •
We propose two hierarchical models to modify the category hierarchy to deal with blocking problems caused by the inconsistency between the artificially defined hierarchy and the actual hierarchy of the data.
- •
We explore three-way decisions to alleviate the classification errors caused by data uncertainty.
Our experiments on the challenging DeepFashion dataset [23] and Stanford Dogs dataset [13] demonstrate the effectiveness of the proposed models. We validate that our TriHC models can significantly reduce the blocking problem when compared with several previous HC models. The classification accuracy can even be higher than that of the well trained deep convolutional neural network model.
Section snippets
Hierarchical classification
In recent years, the development of deep learning has promoted many state-of-the-art models for the classification task [9], [15], [35], [38]. But this kind of flat classification (FC) model wastes the hierarchical relationship between categories. Also when the number of categories is huge, the training of FC models is difficult. The hierarchical classification methods deal with multi-classification problems by dividing a large-scale classification task into several small-scale tasks [8], [12],
Three-way decisions based blocking reduction models
In this section, we first introduce the notation used in this paper. Then, we present two blocking reduction models based on 3WD.
Experiments
In this section, we evaluate the CLPK-TriHC and EC-TriHC on fashion image classification task. Our experiments on the DeepFashion dataset [23] and the Stanford Dogs dataset [13] demonstrate that the proposed methods perform better than several previous HC methods, and even surpass the well trained convolutional neural network (CNN) in some cases.
Conclusion and future works
In this paper, we proposed a three-way decision based hierarchical classification model (TriHC) to alleviate the blocking problem. The TriHC model learns category relations to rebuild the category hierarchy and uses 3WD to targetedly deal with the uncertain data. Adopting different category relation mining methods, we proposed two variants of TriHC, CLPK-TriHC model and EC-TriHC model. Specifically, in CLPK-TriHC model, we considered the cross-level category relationship between the blocked
CRediT authorship contribution statement
Wen Shen: Conceptualization, Investigation, Methodology, Software, Validation, Visualization, Writing - original draft, Writing - review & editing. Zhihua Wei: Supervision. Qianwen Li: Investigation. Hongyun Zhang: Supervision. Duoqian Miao: Supervision.
Declaration of Competing Interest
There is no declaration of interest statement of this paper.
Acknowledgments
The work is partially supported by the National Key Research and Development Project (No. 213), the National Nature Science Foundation of China (No. 61573259, 61976160, 61573255), the Special Project of the Ministry of Public Security (No. 20170004), and the Key Lab of Information Network Security, Ministry of Public Security (No. C18608).
References (50)
- et al.
Hierarchical annotation of medical images
Pattern Recognit.
(2011) - et al.
Soft estimation by hierarchical classification and regression
Neurocomputing
(2017) - et al.
Novel dataset for fine-grained image categorization: Stanford Dogs
Proc. CVPR Workshop on Fine-Grained Visual Categorization (FGVC)
(2011) - et al.
Cost-sensitive sequential three-way decision modeling using a deep neural network
Int. J. Approx. Reason.
(2017) - et al.
Multistage email spam filtering based on three-way decisions
International Conference on Rough Sets and Knowledge Technology
(2013) - et al.
A hierarchical ensemble of ECOC for cancer classification based on multi-class microarray data
Inf Sci
(2016) - et al.
Joint binary classifier learning for ECOC-based multi-class classification
IEEE Trans. Pattern Anal. Mach. Intell.
(2015) Sequential three-way decisions in multi-category image recognition with deep features based on distance factor
Inf. Sci.
(2019)- et al.
A self-adaptive cascade ConvNets model based on three-way decision theory
CCF Chinese Conference on Computer Vision
(2017) - et al.
Hierarchical text classification model based on blocking priori knowledge
Pattern Recognit. Artif. Intell.
(2010)