Applying Cost-Sensitive Multiobjective Genetic Programming to Feature Extraction for Spam E-mail Filtering

* Final gross prices may vary according to local VAT.

Get Access


In this paper we apply multiobjective genetic programming to the cost-sensitive classification task of labelling spam e-mails. We consider three publicly-available spam corpora and make comparison with both support vector machines and naïve Bayes classifiers, both of which are held to perform well on the spam filtering problem. We find that for the high cost ratios of practical interest, our cost-sensitive multiobjective genetic programming gives the best results across a range of performance measures.