论文信息 - Spam Classification Using Adaptive Boosting Algorithm

Spam Classification Using Adaptive Boosting Algorithm

Spam is no doubt a new and growing threat to the Internet and its end users. This paper investigates current approaches for blocking spam and proposes a new spam classification method by using adaptive boosting algorithm. Experiment is carried out to evaluate the results of spam filtering. We find adaptive boosting algorithm is an effective approach to solve the spam problem. We also find that default method in WEKA such as DecisionStump is not actually the best associated algorithm to filter spam. After comparing DecisionStump, J48, and NaiveBayes we conclude J48 is the most suitable associated algorithm to filter spam with high true positive rate, low false positive rate and low computation time.

A. B. M. Shawkat Ali | Yang Xiang | Yang Xiang | A. B. M. S. Ali

[1] Pat Langley,et al. Estimating Continuous Distributions in Bayesian Classifiers , 1995, UAI.

[2] Ian Witten,et al. Data Mining , 2000 .

[3] Gary Robinson,et al. A statistical approach to the spam problem , 2003 .

[4] Yoav Freund,et al. Boosting a weak learning algorithm by majority , 1995, COLT '90.

[5] R. Schapire. The Strength of Weak Learnability , 1990, Machine Learning.

[6] Yoav Freund,et al. A decision-theoretic generalization of on-line learning and an application to boosting , 1995, EuroCOLT.

[7] Yoav Freund,et al. Experiments with a New Boosting Algorithm , 1996, ICML.

[8] Luo Si,et al. A New Boosting Algorithm Using Input-Dependent Regularizer , 2003, ICML 2003.

[9] Dale Schuurmans,et al. Boosting in the Limit: Maximizing the Margin of Learned Ensembles , 1998, AAAI/IAAI.

[10] Yoav Freund,et al. A decision-theoretic generalization of on-line learning and an application to boosting , 1997, EuroCOLT.