On Performance and Calibration of Natural Gradient Langevin Dynamics
Producing deep neural network (DNN) models with calibrated confidence is essential for applications in many fields, such as medical image analysis, natural language processing, and robotics. Modern neural networks have been reported to be poorly calibrated compared with those from a decade ago. The...
Autors principals: | Hanif Amal Robbani, Alhadi Bustamam, Risman Adnan, Shandar Ahmad |
---|---|
Format: | Article |
Idioma: | English |
Publicat: |
IEEE
2023-01-01
|
Col·lecció: | IEEE Access |
Matèries: | |
Accés en línia: | https://ieeexplore.ieee.org/document/10131934/ |
Ítems similars
-
Fast Sampling of Score-Based Models With Cyclical Diffusion Sampling
per: Karimul Makhtidi, et al.
Publicat: (2024-01-01) -
On a generalization of fractional Langevin equation with boundary conditions
per: Zheng Kou, et al.
Publicat: (2022-01-01) -
Dynamical Sampling with Langevin Normalization Flows
per: Minghao Gu, et al.
Publicat: (2019-11-01) -
Calibration with confidence: a principled method for panel assessment
per: R. S. MacKay, et al.
Publicat: (2017-01-01) -
Lévy-walk-like Langevin dynamics
per: Xudong Wang, et al.
Publicat: (2019-01-01)