Construct support vector machine ensemble to detect traffic incident
Expert Systems with Applications: An International Journal
Intelligent detection computer viruses based on multiple classifiers
UIC'07 Proceedings of the 4th international conference on Ubiquitous Intelligence and Computing
Malicious codes detection based on ensemble learning
ATC'07 Proceedings of the 4th international conference on Autonomic and Trusted Computing
Hi-index | 0.00 |
How to generate and aggregate base learners to haveoptimal ensemble generalization capabilities is animportant questions in building compositeregression/classification machines. We present here anevaluation of several algorithms for artificial neuralnetworks aggregation in the regression settings, includingnew proposals and comparing them with standardmethods in the literature. We also discuss a potentialproblem with sequential algorithms: the non-frequent butdamaging selection through their heuristics ofparticularly bad ensemble members. We show that onecan cope with this problem by allowing individualweighting of aggregate members. Our algorithms andtheir weighted modifications are favorably tested againstother methods in the literature, producing a performanceimprovement on the standard statistical databases used asbenchmarks.