neural network theory

Barriers to Discrete Reasoning with Transformers: A Survey Across Depth, Exactness, and Bandwidth EACL 2026 Bayesian Treatment of the Spectrum of the Empirical Kernel in (Sub)Linear-Width Neural Networks ICLR 2025 Decomposition Polyhedra of Piecewise Linear Functions ICLR 2025 Geometry of Lightning Self-Attention: Identifiability and Dimension ICLR 2025 Learning Hierarchical Polynomials of Multiple Nonlinear Features ICLR 2025 A provable control of sensitivity of neural networks through a direct parameterization of the overall bi-Lipschitzness NeurIPS 2024 Learnability of high-dimensional targets by two-parameter models and gradient flow NeurIPS 2024 Mean-Field Analysis for Learning Subspace-Sparse Polynomials with Gaussian Input NeurIPS 2024 On Computational Limits of Modern Hopfield Models: A Fine-Grained Complexity Analysis ICML 2024 On the Power of Small-size Graph Neural Networks for Linear Programming NeurIPS 2024 The effect of Leaky ReLUs on the training and generalization of overparameterized networks AISTATS 2024 Understanding Transformer Reasoning Capabilities via Graph Algorithms NeurIPS 2024 Wide Two-Layer Networks can Learn from Adversarial Perturbations NeurIPS 2024 Backward Feature Correction: How Deep Learning Performs Deep (Hierarchical) Learning COLT 2023 Complex-valued Neurons Can Learn More but Slower than Real-valued Neurons via Gradient Descent NeurIPS 2023 Fundamental Limits of Two-layer Autoencoders, and Achieving Them with Gradient Methods ICML 2023 Gradient Descent in Neural Networks as Sequential Learning in Reproducing Kernel Banach Space ICML 2023 Multi-Layer Neural Networks as Trainable Ladders of Hilbert Spaces ICML 2023 On Bridging the Gap between Mean Field and Finite Width Deep Random Multilayer Perceptron with Batch Normalization ICML 2023 On Single-Index Models beyond Gaussian Data NeurIPS 2023 Towards the Difficulty for a Deep Neural Network to Learn Concepts of Different Complexities NeurIPS 2023 Transformers are uninterpretable with myopic methods: a case study with bounded Dyck grammars NeurIPS 2023 Understanding Multi-phase Optimization Dynamics and Rich Nonlinear Behaviors of ReLU Networks NeurIPS 2023 Width and Depth Limits Commute in Residual Networks ICML 2023 Benefits of Overparameterized Convolutional Residual Networks: Function Approximation under Smoothness Constraint ICML 2022