Generalization

Advances in PAC-Bayesian certification of deep neural networks: tighter closed-form inequalities and optimization of bounds on non-differentiable losses UAI 2026 A Novel Data-Dependent Learning Paradigm for Large Hypothesis Classes ALT 2026 Benign Overfitting with Quantum Kernels UAI 2026 Boosted Control Functions: Distribution Generalization and Invariance in Confounded Models JMLR 2026 Bound to Disagree: Generalization Bounds via Certifiable Surrogates UAI 2026 Bridging Domain Invariance and Diversity: A Fine-Grained Risk Bound for Domain Generalization JMLR 2026 Data Augmentation: A Fourier Analysis Perspective COLT 2026 Echoes Over Time: Unlocking Length Generalization in Video-to-Audio Generation Models CVPR 2026 Exogenous Randomness Empowering Random Forests JMLR 2026 Formalizing Task-Space Complexity for Zero-Shot Generalization L4DC 2026 Is Memorization Helpful or Harmful? Prior Information Sets the Threshold COLT 2026 Margin in Abstract Spaces COLT 2026 On-Average Stability of Multipass Preconditioned SGD and Effective Dimension COLT 2026 Optimal Neural Network Approximation of Smooth Compositional Functions on Sets with Low Intrinsic Dimension COLT 2026 Precise Performance of Linear Denoisers in The Proportional Regime L4DC 2026 Risk Comparisons in Linear Regression: Implicit Regularization Dominates Explicit Regularization (Extended Abstract) COLT 2026 Stochastic Gradient Methods: Bias, Stability and Generalization JMLR 2026 The Distribution of Ridgeless Least Squares Interpolators JMLR 2026 Uniform Convergence Beyond Glivenko-Cantelli ALT 2026 Accelerating Task Generalisation with Multi-Level Skill Hierarchies ICLR 2025 Accuracy on the wrong line: On the pitfalls of noisy data for out-of-distribution generalisation AISTATS 2025 Approximate Equivariance in Reinforcement Learning AISTATS 2025 Asymptotics of Linear Regression with Linearly Dependent Data L4DC 2025 A Theoretical Perspective: How to Prevent Model Collapse in Self-consuming Training Loops ICLR 2025 Black-Box Uniform Stability for Non-Euclidean Empirical Risk Minimization AISTATS 2025