High Probability Convergence of Adam Under Unbounded Gradients and Affine Variance Noise.
Yusu HongJunhong LinPublished in: CoRR (2023)
Keyphrases
- low variance
- low snr
- general regression neural networks
- noise sensitivity
- noise level
- low signal to noise ratio
- noise variance
- high noise
- wide range
- affine transformation
- signal to noise ratio
- noise reduction
- convergence speed
- random noise
- image noise
- noisy environments
- neural network
- noisy data
- image registration
- objective function