Deep double descent: where bigger models and more data hurt*
From MaRDI portal
Publication:5020041
Recommendations
- Double Double Descent: On Generalization Errors in Transfer Learning between Linear Regression Tasks
- Gradient descent optimizes over-parameterized deep ReLU networks
- On deep learning as a remedy for the curse of dimensionality in nonparametric regression
- Machine learning: deepest learning as statistical data assimilation problems
- Deep learning model selection of suboptimal complexity
- Two models of double descent for weak features
- Over-parametrized deep neural networks minimizing the empirical risk do not generalize well
- Why does large batch training result in poor generalization? A comprehensive explanation and a better strategy from the viewpoint of stochastic optimization
- Accelerating deep neural network training with inconsistent stochastic gradient descent
Cites work
Cited in
(36)- On the properties of bias-variance decomposition for kNN regression
- Triple descent and the two kinds of overfitting: where and why do they appear?*
- Sensitivity-informed provable pruning of neural networks
- Binary classification of Gaussian mixtures: abundance of support vectors, benign overfitting, and regularization
- Overparameterization and generalization error: weighted trigonometric interpolation
- A Unifying Tutorial on Approximate Message Passing
- Fit without fear: remarkable mathematical phenomena of deep learning through the prism of interpolation
- Reliable extrapolation of deep neural operators informed by physics or sparse observations
- Consistent Sparse Deep Learning: Theory and Computation
- Semi-Supervised Node Classification via Semi-Global Graph Transformer Based on Homogeneity Augmentation
- Stability of the scattering transform for deformations with minimal regularity
- Fragility, robustness and antifragility in deep learning
- On the robustness of sparse counterfactual explanations to adverse perturbations
- The deep arbitrary polynomial chaos neural network or how deep artificial neural networks could benefit from data-driven homogeneous chaos theory
- Working with machines in mathematics
- A data-driven approach for optimal operational and financial commodity hedging
- Tradeoff of generalization error in unsupervised learning
- Redundant representations help generalization in wide neural networks
- Dropout drops double descent
- Neural networks generalize on low complexity data
- A dynamic view of some anomalous phenomena in SGD for over-parametrized neural networks
- Sharp lower bounds on interpolation by deep ReLU neural networks at irregularly spaced data
- How more data can hurt: instability and regularization in next-generation reservoir computing
- The high-dimensional asymptotics of principal component regression
- A universal law of robustness via isoperimetry
- Learning curves for deep structured Gaussian feature models
- Statistical physics, Bayesian inference, and neural information processing
- Computational complexity of deep learning: fundamental limitations and empirical phenomena
- A primer on Bayesian neural networks: review and debates
- Deep learning generalization and the convex hull of training sets
- Dynamical loss functions shape landscape topography and improve learning in artificial neural networks
- Scaling and renormalization in high-dimensional regression
- Sources of uncertainty in supervised machine learning -- a statisticians' view
- Deep P-Spline: Theory, Fast Tuning, and Application
- The nuclear route: sharp asymptotics of ERM in overparameterized quadratic networks
- On the influence of optimizers in deep learning-based side-channel analysis
This page was built for publication: Deep double descent: where bigger models and more data hurt*
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5020041)
