Results 1 
4 of
4
PolynomialTime Learning with Version Spaces
, 1992
"... Although version spaces provide a useful conceptual tool for inductive concept learning, they often face severe computational difficulties when implemented. For example, the G set of traditional boundaryset implementations of version spaces can have size exponential in the amount of data for even t ..."
Abstract

Cited by 40 (4 self)
 Add to MetaCart
Although version spaces provide a useful conceptual tool for inductive concept learning, they often face severe computational difficulties when implemented. For example, the G set of traditional boundaryset implementations of version spaces can have size exponential in the amount of data for even the most simple conjunctive description languages [ Haussler, 1988 ] . This paper presents a new representation for version spaces that is more general than the traditional boundaryset representation, yet has worstcase time complexity that is polynomial in the amount of data when used for learning from attributevalue data with treestructured feature hierarchies (which includes languages like Haussler's). The central idea underlying this new representation is to maintain the traditional S boundary set as usual, but use a list N of negative data rather than keeping a G set as is typically done. 1. Introduction Concept learning can be viewed as a problem of search [ Simon and Lea, 1974; Mit...
Learning From a Consistently Ignorant Teacher
, 1994
"... One view of computational learning theory is that of a learner acquiring the knowledge of a teacher. We introduce a formal model of learning capturing the idea that teachers may have gaps in their knowledge. In particular, we consider learning from a teacher who labels examples "+" (a positive in ..."
Abstract

Cited by 22 (8 self)
 Add to MetaCart
One view of computational learning theory is that of a learner acquiring the knowledge of a teacher. We introduce a formal model of learning capturing the idea that teachers may have gaps in their knowledge. In particular, we consider learning from a teacher who labels examples "+" (a positive instance of the concept being learned), "\Gamma" (a negative instance of the concept being learned), and "?" (an instance with unknown classification), in such a way that knowledge of the concept class and all the positive and negative examples is not sufficient to determine the labelling of any of the examples labelled with "?". The goal of the learner is not to compensate for the ignorance of the teacher by attempting to infer "+" or "\Gamma" labels for the examples labelled with "?", but is rather to learn (an approximation to) the ternary labelling presented by the teacher. Thus, the goal of the learner is still to acquire the knowledge of the teacher, but now the learner must also ...
The Space of Jumping Emerging Patterns and Its Incremental Maintenance Algorithms
 In Proceedings of the Seventeenth International Conference on Machine Learning
, 2000
"... The concept of jumping emerging patterns (JEPs) has been proposed to describe those discriminating features which only occur in the positive training instances but do not occur in the negative class at all; JEPs have been used to construct classifiers which generally provide better accuracy th ..."
Abstract

Cited by 16 (10 self)
 Add to MetaCart
The concept of jumping emerging patterns (JEPs) has been proposed to describe those discriminating features which only occur in the positive training instances but do not occur in the negative class at all; JEPs have been used to construct classifiers which generally provide better accuracy than the stateoftheart classifiers such as C4.5. The algorithms for maintaining the space of jumping emerging patterns (JEP space) are presented in this paper. We prove that JEP spaces satisfy the property of convexity. Therefore JEP spaces can be concisely represented by two bounds: consisting respectively of the most general elements and the most specific elements. In response to insertion of new training instances, a JEP space is modified by operating on its boundary elements and the boundary elements of the JEP spaces associated with the new instances. This strategy completely avoids the need to go back to the most initial step to build the new JEP space. In addition, our ...