Generalization

Deep Homomorphism Networks (DHNs) boost graph neural network (GNN) expressiveness by efficiently detecting subgraph patterns using a novel graph homomorphism layer.

Credal Learning Theory uses convex sets of probabilities to model data distribution variability, providing theoretical risk bounds for machine learning models in dynamic environments.

New PAC-Bayes bound controls multiple error types simultaneously, providing richer generalization guarantees.

Neural networks, viewed as Bayesian ensembles of fixed classifiers, enable continual learning without forgetting; posterior updates mirror stochastic gradient descent, offering insights into optimizat…

Novel compositional PAC-Bayes framework delivers data-dependent generalization bounds for persistence-enhanced Graph Neural Networks, improving model design and performance.

Neural operators’ continuous discretization is proven impossible in general Hilbert spaces, but achievable using strongly monotone operators, opening new avenues for numerical methods in scientific ma…

New model-agnostic framework for out-of-distribution generalization uses multicalibration across overlapping groups, showing improved robustness and prediction under various distribution shifts.

Leaky ReLU networks exhibit benign overfitting under surprisingly relaxed conditions: input dimension only needs to linearly scale with sample size, challenging prior assumptions in the field.

Despite their brittleness, continuous attractors remain functionally robust analog memory models due to persistent slow manifolds surviving bifurcations, enabling accurate approximation and generaliza…

Many graph neural networks (GNNs) surprisingly converge to constant outputs with increasing graph size, limiting their expressiveness.

Researchers introduce a generalized neural tangent kernel for analyzing surrogate gradient learning in neural networks with non-differentiable activation functions, providing a strong theoretical foun…

This study provides a unified theory for kernel ridge regression’s learning curve, improving existing bounds and validating the Gaussian Equivalence Property under minimal assumptions.