Fisher information-based metrics for representation learning
Published in 2025 Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), 2025
This paper proposes Fisher information-based metrics for representation learning in deep neural networks. The proposed metrics provide principled measures that leverage the geometry of the model parameter space to guide and improve the quality of learned representations.
Recommended citation: Do Nguyen Dang Thi, Le Quoc Anh, Tran Trong Duy, Le Vu Ha, and Nguyen Linh Trung. "Fisher information-based metrics for representation learning." In 2025 Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), pages 1556–1561, 2025.
Download Paper
