Missing data is a common problem in credit evaluation practice and can obstruct the development and application of an evaluation model. Block-wise missing data is a particularly troublesome issue. Based on multi-task feature selection approach, this paper proposes a method called MMPFS to build a model for credit evaluation that primarily includes two steps: (1) dividing the dataset into several nonoverlapping subsets based on missing patterns, and (2) integrating the multi-task feature selection approach using logistic regression to perform joint feature learning on all subsets. The proposed method has the following advantages: (1) missing data do not need to be managed in advance, (2) available data can be fully used for model learning, (3) information loss or bias caused by general missing data processing methods can be avoided, and (4) overfitting risk caused by redundant features can be reduced. The implementation framework and algorithm principle of the proposed method are described, and three credit datasets from UCI are investigated to compare the proposed method with other commonly used missing data treatments. The results show that MMPFS can produce a better credit evaluation model than data preprocessing methods, such as sample deletion and data imputation.
In this paper, on the basis of breadth-first and depth-first ways, we establish a fundamental framework of fuzzy grammars based on lattices, which provides a necessary tool for the analysis of fuzzy automata. The relationship among finite automata with membership values in lattices (l-VFAs), lattice-valued regular grammars (l-RGs) and lattice-valued deterministic regular grammars (l-DRGs) is investigated. It is demonstrated that, based on each semantic way, l-VFAs and l-RGs are equivalent in the sense that they accept or generate the same classes of fuzzy languages. Furthermore, it is proved that l-VFAs,?l-valued deterministic finite automata, l-RGs and l-DRGs are equivalent based on depth-first way. For any l-RG,?the language based on breadth-first way coincides with the language based on depth-first way if and only if the truth-valued domain l is a distributive lattice. 相似文献
This paper presents a novel online learning visual servo controller integrating the FCMAC with proportion controller for the
control of position of manipulator end-effector. Since the FCMAC has good learning capability and fast learning speed, and
can save much computer memory space by fuzzy processing of input space division and memory unit activation, it is used to
develop an adaptive control law by learning the relationship between the image feature errors and manipulator input, and the
aim of online learning of the FCMAC is to minimize the output of proportion controller. Furthermore, the FCMAC has no need
for models of robot manipulator and image feature extraction, so that the capability of proposed controller for tasks under
uncertain environment can be improved. Finally, the proposed controller is proved to be effective by the experiment, and compared
with BP neural network. 相似文献