overparameterization
The condition where a model has more parameters than necessary, which can lead to better training performance but may also raise concerns about overfitting and generalization.
- Composing Global Solutions to Reasoning Tasks via Algebraic Objects in Neural Nets
- Dependency Parsing is More Parameter-Efficient with Normalization
- Differentiable Sparsity via $D$-Gating: Simple and Versatile Structured Penalization
- Double Descent Meets Out-of-Distribution Detection: Theoretical Insights and Empirical Analysis on the Role of Model Complexity
- Guarantees for Alternating Least Squares in Overparameterized Tensor Decompositions
- Light-Weight Diffusion Multiplier and Uncertainty Quantification for Fourier Neural Operators
- RGNMR: A Gauss-Newton method for robust matrix completion with theoretical guarantees
- Self-Assembling Graph Perceptrons
- Solving Neural Min-Max Games: The Role of Architecture, Initialization & Dynamics