Research map / Learning theory and algorithms / Kernels and regression theory
Tilting the playing field: Dynamical loss functions for machine learning
Miguel Ruiz-Garcia, Ge Zhang, Samuel S Schoenholz, Andrea J. Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Cycle class weights during training (A∼50, T∼100) | Myrtle5, 64 channels · CIFAR10 validation | accuracy: ~ 0.73 -> ~ 0.79; improves up to 6% | Improved |
| Cycle class weights (5≲A≲20, T≲300) | One-hidden-layer network, width 100 · Spiral dataset | training accuracy: ~ 0.65 -> nearly perfect | Improved |
| Cycle class weights (5≲A≲20, T≲300) | One-hidden-layer network, width 100 · Spiral dataset | validation accuracy: ~ 0.65 -> ~ 0.9 | Improved |
| Cycle class weights (5≲A≲25, T≲700) | One-hidden-layer network, width 1000 · Spiral dataset | generalization accuracy | Improved |
| Apply oscillatory loss weights | Wide Residual Network · CIFAR10 test | accuracy | No clear effect |
| Increase oscillation amplitude beyond the optimum (A≈30) (ablation) | Dynamical loss · Classification tasks | accuracy | Hurt |
Closest papers by meaning
- Dynamical Decoupling of Generalization and Overfitting in Large Two-Layer Networks (NeurIPS 2025 · similarity 0.89)
- Visualizing the Loss Landscape of Neural Nets (NeurIPS 2018 · similarity 0.89)
- Taxonomizing local versus global structure in neural network loss landscapes (NeurIPS 2021 · similarity 0.88)
- Loss Landscape Characterization of Neural Networks without Over-Parametrization (NeurIPS 2024 · similarity 0.88)
- Learning Dynamics of Deep Matrix Factorization Beyond the Edge of Stability (ICLR 2025 · similarity 0.88)
- Deep Linear Network Training Dynamics from Random Initialization: Data, Width, Depth, and Hyperparameter Transfer (ICML 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
