% Key papers for Generalization theory of deep networks (scime atlas, https://atlas.scime.ai). Authors are not included: add them from the paper pages.

@inproceedings{neurips2017_spectrally_56680,
  title = {{Spectrally-normalized margin bounds for neural networks}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2017},
  url = {https://nips.cc/virtual/2017/poster/9395},
}

@inproceedings{neurips2018_entropy_57099,
  title = {{Entropy and mutual information in models of deep neural networks}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2018},
  url = {https://nips.cc/virtual/2018/poster/11195},
}

@inproceedings{neurips2021_a_63599,
  title = {{A Universal Law of Robustness via Isoperimetry}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2021},
  url = {https://nips.cc/virtual/2021/poster/27813},
}

@inproceedings{neurips2019_random_58877,
  title = {{Random deep neural networks are biased towards simple functions}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2019},
  url = {https://nips.cc/virtual/2019/poster/13401},
}

@inproceedings{neurips2018_the_57701,
  title = {{The committee machine: Computational to statistical gaps in learning a two-layers neural network}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2018},
  url = {https://nips.cc/virtual/2018/poster/11326},
}

@inproceedings{neurips2017_net_56506,
  title = {{Net-Trim: Convex Pruning of Deep Neural Networks with Performance Guarantee}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2017},
  url = {https://nips.cc/virtual/2017/poster/9102},
}

@inproceedings{aaai2021_on_1088,
  title = {{On Lipschitz Regularization of Convolutional Layers using Toeplitz Matrix Theory}},
  booktitle = {AAAI Conference on Artificial Intelligence (AAAI)},
  year = {2021},
  url = {https://aaai.org/papers/06661-on-lipschitz-regularization-of-convolutional-layers-using-toeplitz-matrix-theory/},
}

@inproceedings{iclr2018_adaptive_28590,
  title = {{Adaptive Dropout with Rademacher Complexity Regularization}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2018},
  url = {https://iclr.cc/virtual/2018/poster/287},
}

@inproceedings{iclr2020_truth_29573,
  title = {{Truth or backpropaganda? An empirical investigation of deep learning theory}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2020},
  url = {https://openreview.net/forum?id=HyxyIgHFvr},
}

@inproceedings{icml2023_how_42640,
  title = {{How Does Information Bottleneck Help Deep Learning?}},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2023},
  url = {https://icml.cc/virtual/2023/poster/25178},
}

@inproceedings{icml2023_a_42777,
  title = {{A Toy Model of Universality: Reverse Engineering how Networks Learn Group Operations}},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2023},
  url = {https://icml.cc/virtual/2023/poster/24596},
}

@inproceedings{aaai2019_capacity_101828,
  title = {{Capacity Control of ReLU Neural Networks by Basis-Path Norm}},
  booktitle = {AAAI Conference on Artificial Intelligence (AAAI)},
  year = {2019},
  url = {https://ojs.aaai.org/index.php/AAAI/article/view/4543},
}

@inproceedings{neurips2018_understanding_57739,
  title = {{Understanding Weight Normalized Deep Neural Networks with Rectified Linear Units}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2018},
  url = {https://nips.cc/virtual/2018/poster/11040},
}

@inproceedings{iclr2022_machine_31647,
  title = {{Machine Learning For Elliptic PDEs: Fast Rate Generalization Bound, Neural Scaling Law and Minimax Optimality}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2022},
  url = {https://iclr.cc/virtual/2022/poster/6547},
}

@inproceedings{iclr2020_beyond_29860,
  title = {{Beyond Linearization: On Quadratic and Higher-Order Approximation of Wide Neural Networks}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2020},
  url = {https://openreview.net/forum?id=rkllGyBFPH},
}

@inproceedings{iclr2021_estimating_30374,
  title = {{Estimating Lipschitz constants of monotone deep equilibrium models}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2021},
  url = {https://iclr.cc/virtual/2021/poster/3308},
}

@inproceedings{icml2021_tilting_40034,
  title = {{Tilting the playing field: Dynamical loss functions for machine learning}},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2021},
  url = {https://icml.cc/virtual/2021/poster/9505},
}

@inproceedings{neurips2020_analytic_59383,
  title = {{Analytic Characterization of the Hessian in Shallow ReLU Models: A Tale of Symmetry}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2020},
  url = {https://nips.cc/virtual/2020/poster/17303},
}

@inproceedings{neurips2021_relative_62214,
  title = {{Relative stability toward diffeomorphisms indicates performance in deep nets}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2021},
  url = {https://nips.cc/virtual/2021/poster/28781},
}

@inproceedings{iclr2019_data_28907,
  title = {{Data-Dependent Coresets for Compressing Neural Networks with Applications to Generalization Bounds}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2019},
  url = {https://iclr.cc/virtual/2019/poster/868},
}

@inproceedings{icml2024_deeper_44027,
  title = {{Deeper or Wider: A Perspective from Optimal Generalization Error with Sobolev Loss}},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2024},
  url = {https://icml.cc/virtual/2024/poster/34425},
}

@inproceedings{aaai2025_optimal_10211,
  title = {{Optimal Control Operator Perspective and a Neural Adaptive Spectral Method}},
  booktitle = {AAAI Conference on Artificial Intelligence (AAAI)},
  year = {2025},
  url = {https://ojs.aaai.org/index.php/AAAI/article/view/33596},
}

@inproceedings{iclr2024_quadratic_34386,
  title = {{Quadratic models for understanding catapult dynamics of neural networks}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2024},
  url = {https://iclr.cc/virtual/2024/poster/18689},
}

@inproceedings{neurips2024_optimal_74343,
  title = {{Optimal deep learning of holomorphic operators between Banach spaces}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2024},
  url = {https://neurips.cc/virtual/2024/poster/93217},
}

@inproceedings{neurips2024_topological_70594,
  title = {{Topological Generalization Bounds for Discrete-Time Stochastic Optimization Algorithms}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2024},
  url = {https://neurips.cc/virtual/2024/poster/96523},
}

@inproceedings{neurips2024_a_72317,
  title = {{A two-scale Complexity Measure for Deep Learning Models}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2024},
  url = {https://neurips.cc/virtual/2024/poster/95029},
}

@inproceedings{neurips2025_learning_77078,
  title = {{Learning Linear Attention in Polynomial Time}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2025},
  url = {https://openreview.net/forum?id=QN0E0KX2LM},
}

@inproceedings{iclr2024_a_34985,
  title = {{A path-norm toolkit for modern networks: consequences, promises and challenges}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2024},
  url = {https://iclr.cc/virtual/2024/poster/18089},
}

@inproceedings{iclr2024_feature_35001,
  title = {{Feature emergence via margin maximization: case studies in algebraic tasks}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2024},
  url = {https://iclr.cc/virtual/2024/poster/18073},
}

@inproceedings{iclr2024_generalization_35454,
  title = {{Generalization of Scaled Deep ResNets in the Mean-Field Regime}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2024},
  url = {https://iclr.cc/virtual/2024/poster/17619},
}

@inproceedings{icml2024_exploring_44380,
  title = {{Exploring the Complexity of Deep Neural Networks through Functional Equivalence}},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2024},
  url = {https://icml.cc/virtual/2024/poster/34072},
}

@inproceedings{icml2024_subhomogeneous_44671,
  title = {{Subhomogeneous Deep Equilibrium Models}},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2024},
  url = {https://icml.cc/virtual/2024/poster/33781},
}

@inproceedings{neurips2025_dynamical_76398,
  title = {{Dynamical Decoupling of Generalization and Overfitting in Large Two-Layer Networks}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2025},
  url = {https://openreview.net/forum?id=ImpizBSKcu},
}

@inproceedings{neurips2025_on_77539,
  title = {{On Universality Classes of Equivariant Networks}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2025},
  url = {https://openreview.net/forum?id=V4YAS7NLXi},
}

@inproceedings{iclr2025_generalization_81894,
  title = {{Generalization in VAE and Diffusion Models: A Unified Information-Theoretic Analysis}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2025},
  url = {https://iclr.cc/virtual/2025/poster/29890},
}

@inproceedings{iclr2025_how_82351,
  title = {{How DNNs break the Curse of Dimensionality: Compositionality and Symmetry Learning}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2025},
  url = {https://iclr.cc/virtual/2025/poster/29440},
}

@inproceedings{iclr2026_bridging_85854,
  title = {{Bridging Kolmogorov Complexity and Deep Learning: Asymptotically Optimal Description Length Objectives for Transformers}},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2026},
  url = {https://openreview.net/forum?id=IwX6Gcvt0j},
}

@inproceedings{neurips2026_generalization_95437,
  title = {{Generalization at the Edge of Stability}},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2026},
  url = {https://neurips.cc/virtual/2026/poster/154976},
}

@inproceedings{icml2025_position_46371,
  title = {{Position: A Theory of Deep Learning Must Include Compositional Sparsity}},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2025},
  url = {https://icml.cc/virtual/2025/poster/40171},
}

@inproceedings{icml2026_biologically_49747,
  title = {{Biologically plausible heavy-tailed connectivity enhances generalizations on cognitive tasks in recurrent neural networks}},
  booktitle = {International Conference on Machine Learning (ICML)},
  year = {2026},
  url = {https://openreview.net/forum?id=5Mt4Jdrqfz},
}