Research On Multi-label Learning Algorithms With Labeling Information Enrichment

Multi-label learning deals with training examples each represented by a single instance(feature vector)while associated with multiple class labels,and the task is to train a predictive model which can assign a set of proper labels for the unseen instance.Previous studies show that multi-label learning can learn from real-world objects with rich semantics effectively.In the past decade,multi-label learning has attracted great interests and numerous effective learning methods have been proposed.Nonetheless,existing approaches learn from multi-label data by utilizing the original binary labeling information,while the rich semantics of multi-label data may not be fully exploited.In this thesis,the problem of multi-label learning with labeling information enrichment is studied:On the one hand,existing approaches learn from multi-label examples by taking the common assumption of equal labeling-importance.Nonetheless,this common assumption does not reflect the fact that the importance degree of each relevant label is generally different,though the importance information is not directly accessible from the training examples.In this thesis,we show that it is beneficial to leverage the implicit relative labeling-importance(RLI)information to help induce multi-label predictive model with strong generalization performance.Accordingly,a novel multi-label learning approach named Reliab(RElative Labeling-Importance Aware multi-laBel learning)is proposed.Specifically,RLI degrees are formalized as multinomial distribution over the label space,which can be estimated by either global label propagation procedure or local k-nearest neighbor reconstruction.Correspondingly,the multi-label predictive model is induced by fitting modeling outputs with estimated RLI degrees along with multi-label empirical loss regularization.Extensive experiments clearly validate that leveraging implicit RLI information serves as a favorable strategy to achieve effective multi-label learning.On the other hand,most existing approaches make use of multi-label training examples by exploiting their labeling information in a crisp manner,i.e.one class label is either fully relevant or irrelevant to the instance.In this thesis,a novel multi-label learning approach named Mlfe(Multi-label Learning with Feature-induced labeling information enrichment)is proposed which aims to enrich the labeling information by leveraging the structural information in feature space.Firstly,the underlying structure of feature space is characterized by conducting sparse reconstruction among the training examples.Secondly,the reconstruction information is conveyed from feature space to label space so as to enrich the original categorical labels into numerical ones.Thirdly,the multi-label predictive model is induced by learning from training examples with enriched labeling information.Extensive experiments clearly validate the effectiveness of the proposed feature-induced strategy for enhancing labeling information of multi-label examples.There are five chapters in this thesis.In Chapter 1,we give a brief overview of multi-label learning research and the problems remaining to be studied.In Chapter 2,formal definition on multi-label learning is given and several representative multi-label learning algorithms are introduced.Chapter 3 and 4 respectively introduce two kinds of multi-label learning algorithm based on labeling information enrichment,including Reliab and Mlfe.In Chapter 5,we summarize the whole thesis and discuss several issues for future work.