The Optimized Anomaly Detection Models Based on an Approach of Dealing with Imbalanced Dataset for Credit Card Fraud Detection

Algorithm 4

OCSVM algorithm with AdaBoost [16]

Input

 Training set:

 The probability distribution of each data:

 Maximum number of iterations:

Output

 , which is the ensemble of the base classifiers.

Initialize:

For

 Step 1: train base learner by distribution .

 Step 2: get base classifiers , get the rate of the error-classified samples

 Step 3 choose is a constant.

 Step 4: update

  where is a normalization factor.

End for

Output the final hypothesis: