Multi-label learning deals with data associated with a set of labels simultaneously. Dimensionality reduction is an important but challenging task in multi-label learning. Feature selection is an efficient technique f...Multi-label learning deals with data associated with a set of labels simultaneously. Dimensionality reduction is an important but challenging task in multi-label learning. Feature selection is an efficient technique for dimensionality reduction to search an optimal feature subset preserving the most relevant information. In this paper, we propose an effective feature evaluation criterion for multi-label feature selection, called neighborhood relationship preserving score. This criterion is inspired by similarity preservation, which is widely used in single-label feature selection. It evaluates each feature subset by measuring its capability in preserving neighborhood relationship among samples. Unlike similarity preservation, we address the order of sample similarities which can well express the neighborhood relationship among samples, not just the pairwise sample similarity. With this criterion, we also design one ranking algorithm and one greedy algorithm for feature selection problem. The proposed algorithms are validated in six publicly available data sets from machine learning repository. Experimental results demonstrate their superiorities over the compared state-of-the-art methods.展开更多
In many machine learning applications,data are not free,and there is a test cost for each data item. For the economical reason,some existing works try to minimize the test cost and at the same time,preserve a particul...In many machine learning applications,data are not free,and there is a test cost for each data item. For the economical reason,some existing works try to minimize the test cost and at the same time,preserve a particular property of a given decision system. In this paper,we point out that the test cost one can afford is limited in some applications. Hence,one has to sacrifice respective properties to keep the test cost under a budget. To formalize this issue,we define the test cost constraint attribute reduction problem,where the optimization objective is to minimize the conditional information entropy. This problem is an essential generalization of both the test-cost-sensitive attribute reduction problem and the 0-1 knapsack problem,therefore it is more challenging. We propose a heuristic algorithm based on the information gain and test costs to deal with the new problem. The algorithm is tested on four UCI(University of California-Irvine) datasets with various test cost settings. Experimental results indicate the appropriate setting of the only user-specified parameter λ.展开更多
Rough set theory is a technique of granular computing. In this paper, we study a type of generalized rough sets based on covering. There are several literatures[1,40-43] exploring covering-based rough sets. Our focus ...Rough set theory is a technique of granular computing. In this paper, we study a type of generalized rough sets based on covering. There are several literatures[1,40-43] exploring covering-based rough sets. Our focus of this paper is on the dualities in rough operations.展开更多
基金supported in part by the National Natural Science Foundation of China(61379049,61772120)
文摘Multi-label learning deals with data associated with a set of labels simultaneously. Dimensionality reduction is an important but challenging task in multi-label learning. Feature selection is an efficient technique for dimensionality reduction to search an optimal feature subset preserving the most relevant information. In this paper, we propose an effective feature evaluation criterion for multi-label feature selection, called neighborhood relationship preserving score. This criterion is inspired by similarity preservation, which is widely used in single-label feature selection. It evaluates each feature subset by measuring its capability in preserving neighborhood relationship among samples. Unlike similarity preservation, we address the order of sample similarities which can well express the neighborhood relationship among samples, not just the pairwise sample similarity. With this criterion, we also design one ranking algorithm and one greedy algorithm for feature selection problem. The proposed algorithms are validated in six publicly available data sets from machine learning repository. Experimental results demonstrate their superiorities over the compared state-of-the-art methods.
基金supported by the National Natural Science Foundation of China under Grant No. 60873077/F020107
文摘In many machine learning applications,data are not free,and there is a test cost for each data item. For the economical reason,some existing works try to minimize the test cost and at the same time,preserve a particular property of a given decision system. In this paper,we point out that the test cost one can afford is limited in some applications. Hence,one has to sacrifice respective properties to keep the test cost under a budget. To formalize this issue,we define the test cost constraint attribute reduction problem,where the optimization objective is to minimize the conditional information entropy. This problem is an essential generalization of both the test-cost-sensitive attribute reduction problem and the 0-1 knapsack problem,therefore it is more challenging. We propose a heuristic algorithm based on the information gain and test costs to deal with the new problem. The algorithm is tested on four UCI(University of California-Irvine) datasets with various test cost settings. Experimental results indicate the appropriate setting of the only user-specified parameter λ.
文摘Rough set theory is a technique of granular computing. In this paper, we study a type of generalized rough sets based on covering. There are several literatures[1,40-43] exploring covering-based rough sets. Our focus of this paper is on the dualities in rough operations.