A genetic programming method for classifier construction and cost learning in high-dimensional unbalanced classification
Wenbin Pei, Bing Xue, Lin Shang, Mengjie Zhang · 2020
Cost-sensitive learning has been widely used to address the problem of class imbalance. However, cost matrices are often manually designed. In many real-world applications, cost values are often unknown because of the limited domain knowledge. This paper proposes a new genetic programming method to construct cost-sensitive classifiers, which do not require the manually designed cost values. The experimental results show that the proposed method often outperforms existing GP methods.