spectral bias
Spectral bias refers to the tendency of certain models, particularly neural networks, to learn low-frequency components of a signal faster than high-frequency components. Understanding spectral bias is key for improving how models generalize across different complexities of data.
- An Analytical Theory of Spectral Bias in the Learning Dynamics of Diffusion Models
- FAPEX: Fractional Amplitude-Phase Expressor for Robust Cross-Subject Seizure Prediction
- Hybrid Autoencoders for Tabular Data: Leveraging Model-Based Augmentation in Low-Label Settings
- Neuro-Spectral Architectures for Causal Physics-Informed Networks
- PDEfuncta: Spectrally-Aware Neural Representation for PDE Solution Modeling
- SHAP values via sparse Fourier representation
- Self-diffusion for Solving Inverse Problems