THE CAUCHY-SCHWARZ DIVERGENCE AND ENTROPY-BASED DENSITY-WEIGHTED ACTIVE LEARNING


Creative Commons License

Thesis Type: Postgraduate

Institution Of The Thesis: Middle East Technical University, Graduate School of Natural and Applied Sciences, Graduate School of Natural and Applied Sciences, Turkey

Approval Date: 2022

Thesis Language: English

Student: MEHMET ENES ŞEN

Supervisor: Barış Nakiboğlu

Abstract:

The general framework for active learning is explained. The existing active learning strategies are surveyed. The information-theoretic measures such as the entropy and the mutual information are analyzed as active learning objectives. The use of divergence measures in density-weighted active learning is discussed. A novel density-weighted active learning algorithm, based on Cauchy-Schwarz divergence and entropy, is introduced and compared with the state-of-the-art active learning strategies.