Author(s): Bhoomi Trivedi | Neha Kapadia
Journal: International Journal of Computer Applications
ISSN 0975-8887
Volume: icwet;
Issue: 13;
Date: 2012;
Original page
Keywords: Stacked Generalization | sequential stacked generalization | ensemble learning | multiple classifier system
ABSTRACT
Nowadays machine learning techniques can be successfully applied to data mining tasks. In inductive machine learning, combination of several classifiers is very lively field and has shown favorable results compare to those of single expert systems for variety of scenarios. In this paper one of the ensemble learning method, i.e stacked generalization is modified to get better predictive accuracy. In stacking, by knowing its area of expertise, different diverse base classifiers are combined by a learnable combiner. So error can be generalized by the combiner. As diversity is the important aspect of the ensemble learning, in this paper sequential learning of the base classifier is experimented for that. To evaluate the performance of the proposed method different data sets like, IONOSPHERE, HYPOTHYROID, WAVEFORM are used. The experiments demonstrate the efficiency of the proposed model in terms of accuracy and time by yielding higher accuracy and lesser time relative to conventional staked generalization method
Journal: International Journal of Computer Applications
ISSN 0975-8887
Volume: icwet;
Issue: 13;
Date: 2012;
Original page
Keywords: Stacked Generalization | sequential stacked generalization | ensemble learning | multiple classifier system
ABSTRACT
Nowadays machine learning techniques can be successfully applied to data mining tasks. In inductive machine learning, combination of several classifiers is very lively field and has shown favorable results compare to those of single expert systems for variety of scenarios. In this paper one of the ensemble learning method, i.e stacked generalization is modified to get better predictive accuracy. In stacking, by knowing its area of expertise, different diverse base classifiers are combined by a learnable combiner. So error can be generalized by the combiner. As diversity is the important aspect of the ensemble learning, in this paper sequential learning of the base classifier is experimented for that. To evaluate the performance of the proposed method different data sets like, IONOSPHERE, HYPOTHYROID, WAVEFORM are used. The experiments demonstrate the efficiency of the proposed model in terms of accuracy and time by yielding higher accuracy and lesser time relative to conventional staked generalization method
