Theory

Towards Better Understanding Attribution Methods CVPR 2022 Towards Bridging Sample Complexity and Model Capacity AAAI 2022 Towards Theoretical Analysis of Transformation Complexity of ReLU DNNs ICML 2022 Towards Understanding Grokking: An Effective Theory of Representation Learning NeurIPS 2022 Towards Understanding the Mixture-of-Experts Layer in Deep Learning NeurIPS 2022 Training Scale-Invariant Neural Networks on the Sphere Can Happen in Three Regimes NeurIPS 2022 Training Two-Layer ReLU Networks with Gradient Descent is Inconsistent JMLR 2022 Transformers from an Optimization Perspective NeurIPS 2022 Transition to Linearity of General Neural Networks with Directed Acyclic Graph Architecture NeurIPS 2022 Truncation Sampling as Language Model Desmoothing EMNLP 2022 Two-layer neural network on infinite dimensional data: global optimization guarantee in the mean-field regime NeurIPS 2022 Uncertainty Modeling in Generative Compressed Sensing ICML 2022 Understanding and Leveraging Overparameterization in Recursive Value Estimation ICLR 2022 Understanding Square Loss in Training Overparametrized Neural Network Classifiers NeurIPS 2022 Understanding the Generalization Benefit of Normalization Layers: Sharpness Reduction NeurIPS 2022 Understanding Uncertainty Maps in Vision With Statistical Testing CVPR 2022 Universality of Group Convolutional Neural Networks Based on Ridgelet Analysis on Groups NeurIPS 2022 Universal Joint Approximation of Manifolds and Densities by Simple Injective Flows ICML 2022 Vanishing Curvature in Randomly Initialized Deep ReLU Networks AISTATS 2022 Verification of Neural-Network Control Systems by Integrating Taylor Models and Zonotopes AAAI 2022 What Can the Neural Tangent Kernel Tell Us About Adversarial Robustness? NeurIPS 2022 What Happens after SGD Reaches Zero Loss? --A Mathematical Framework ICLR 2022 What You See is What You Get: Principled Deep Learning via Distributional Generalization NeurIPS 2022 Where do Models go Wrong? Parameter-Space Saliency Maps for Explainability NeurIPS 2022 Whitening Convergence Rate of Coupling-based Normalizing Flows NeurIPS 2022