Theory

Phase Transition from Clean Training to Adversarial Training NeurIPS 2022 Post-Hoc Interpretation of Transformer Hyperparameters with Explainable Boosting Machines EMNLP 2022 Principal Components Bias in Over-parameterized Linear Models, and its Manifestation in Deep Neural Networks JMLR 2022 Probing for the Usage of Grammatical Number ACL 2022 Prompt Certified Machine Unlearning with Randomized Gradient Smoothing and Quantization NeurIPS 2022 Provable Guarantees for Understanding Out-of-Distribution Detection AAAI 2022 Rank Diminishing in Deep Neural Networks NeurIPS 2022 Real-Valued Backpropagation is Unsuitable for Complex-Valued Neural Networks NeurIPS 2022 Redundant representations help generalization in wide neural networks NeurIPS 2022 Rethinking Influence Functions of Neural Networks in the Over-Parameterized Regime AAAI 2022 Rethinking InfoNCE: How Many Negative Samples Do You Need? IJCAI 2022 Rethinking Lipschitz Neural Networks and Certified Robustness: A Boolean Function Perspective NeurIPS 2022 Reverse Engineering the Neural Tangent Kernel ICML 2022 Revisiting Over-smoothing in BERT from the Perspective of Graph ICLR 2022 RL with KL penalties is better viewed as Bayesian inference EMNLP 2022 Robust Fine-Tuning of Deep Neural Networks with Hessian-based Generalization Guarantees ICML 2022 Robustness Implies Generalization via Data-Dependent Generalization Bounds ICML 2022 Robustness in deep learning: The good (width), the bad (depth), and the ugly (initialization) NeurIPS 2022 Safety Guarantees for Neural Network Dynamic Systems via Stochastic Barrier Functions NeurIPS 2022 Scale Mixtures of Neural Network Gaussian Processes ICLR 2022 Self-Consistent Dynamical Field Theory of Kernel Evolution in Wide Neural Networks NeurIPS 2022 Sharp Analysis of Random Fourier Features in Classification AAAI 2022 Signal Processing for Implicit Neural Representations NeurIPS 2022 Signal Propagation in Transformers: Theoretical Perspectives and the Role of Rank Collapse NeurIPS 2022 Signal Recovery with Non-Expansive Generative Network Priors NeurIPS 2022