Learning class-imbalanced problems from the perspective of data intrinsic characteristics

Kong, J.

Persistent URL of this record https://hdl.handle.net/1887/3642254

View statistics

In Collections

This item can be found in the following collections:

Dissertations

Kong, J. (2023)

Learning class-imbalanced problems from the perspective of data intrinsic characteristics

Doctoral Thesis

The class-imbalance problem is a challenging classification task and is frequently encountered in real-world applications. Various techniques have been developed to improve the imbalanced classification performance theoretically and practically. Apart from developing new approaches, researchers also address the importance of understanding the data itself, which will provide more insight into what actually hinders the imbalanced classification performance. This thesis mainly conducts the research on Learning Class-Imbalanced Problems from the Perspective of Data Intrinsic Characteristics.

Data Complexity

Anomaly Detection

Resampling Techniques

Hyperparameter Optimisation

Imbalanced Application