{"total":443,"rows":[{"id":"2209.04836","title":"Git Re-Basin: Merging Models modulo Permutation Symmetries","created":"2022-09-11","cited_by_count":471,"area":null},{"id":"2110.06296","title":"The Role of Permutation Invariance in Linear Mode Connectivity of Neural Networks","created":"2021-10-12","cited_by_count":291,"area":null},{"id":"2101.12699","title":"Exploring Deep Neural Networks via Layer-Peeled Model: Minority Collapse in Imbalanced Training","created":"2021-01-29","cited_by_count":166,"area":null},{"id":"2312.10794","title":"A mathematical perspective on Transformers","created":"2023-12-17","cited_by_count":164,"area":null},{"id":"2206.15144","title":"Neural Networks can Learn Representations with Gradient Descent","created":"2022-06-30","cited_by_count":161,"area":null},{"id":"2106.02073","title":"Neural Collapse Under MSE Loss: Proximity to and Dynamics on the Central Path","created":"2021-06-03","cited_by_count":158,"area":null},{"id":"2205.01445","title":"High-dimensional Asymptotics of Feature Learning: How One Gradient Step Improves the Representation","created":"2022-05-03","cited_by_count":150,"area":null},{"id":"2302.11055","title":"SGD learning on neural networks: leap complexity and saddle-to-saddle dynamics","created":"2023-02-21","cited_by_count":143,"area":null},{"id":"2110.11773","title":"Sinkformers: Transformers with Doubly Stochastic Attention","created":"2021-10-22","cited_by_count":138,"area":null},{"id":"2211.08403","title":"REPAIR: REnormalizing Permuted Activations for Interpolation Repair","created":"2022-11-15","cited_by_count":120,"area":null},{"id":"2106.09524","title":"Implicit Bias of SGD for Diagonal Linear Networks: a Provable Benefit of Stochasticity","created":"2021-06-17","cited_by_count":110,"area":null},{"id":"2310.06110","title":"Grokking as the Transition from Lazy to Rich Training Dynamics","created":"2023-10-09","cited_by_count":108,"area":null},{"id":"2205.09653","title":"Self-Consistent Dynamical Field Theory of Kernel Evolution in Wide Neural Networks","created":"2022-05-19","cited_by_count":105,"area":null},{"id":"2107.13163","title":"Statistically Meaningful Approximation: a Case Study on Approximating Turing Machines with Transformers","created":"2021-07-28","cited_by_count":102,"area":null},{"id":"2305.16380","title":"Scan and Snap: Understanding Training Dynamics and Token Composition in 1-layer Transformer","created":"2023-05-25","cited_by_count":102,"area":null},{"id":"2203.09081","title":"Inducing Neural Collapse in Imbalanced Learning: Do We Really Need a Learnable Classifier at the End of Deep Neural Network?","created":"2022-03-17","cited_by_count":100,"area":null},{"id":"2106.15933","title":"Saddle-to-Saddle Dynamics in Deep Linear Networks: Small Initialization Training, Symmetry, and Sparsity","created":"2021-06-30","cited_by_count":85,"area":null},{"id":"2301.02679","title":"Grokking modular arithmetic","created":"2023-01-06","cited_by_count":83,"area":null},{"id":"2106.15013","title":"Small random initialization is akin to spectral learning: Optimization and generalization guarantees for overparameterized low-rank matrix reconstruction","created":"2021-06-28","cited_by_count":78,"area":null},{"id":"2505.01618","title":"Don't be lazy: CompleteP enables compute-efficient deep transformers","created":"2025-05-02","cited_by_count":78,"area":null},{"id":"2206.00939","title":"Gradient flow dynamics of shallow ReLU networks for square loss and orthogonal inputs","created":"2022-06-02","cited_by_count":76,"area":null},{"id":"2103.01550","title":"Label-Imbalanced and Group-Sensitive Classification under Overparameterization","created":"2021-03-02","cited_by_count":74,"area":null},{"id":"2301.13105","title":"Generalization on the Unseen, Logic Reasoning and Degree Curriculum","created":"2023-01-30","cited_by_count":73,"area":null},{"id":"2206.04030","title":"High-dimensional limit theorems for SGD: Effective dynamics and critical scaling","created":"2022-06-08","cited_by_count":70,"area":null},{"id":"2102.11742","title":"Classifying high-dimensional Gaussian mixtures: Where kernel methods fail and neural networks succeed","created":"2021-02-23","cited_by_count":69,"area":null},{"id":"2201.07395","title":"Overview frequency principle/spectral bias in deep learning","created":"2022-01-19","cited_by_count":69,"area":null},{"id":"2305.18270","title":"How Two-Layer Neural Networks Learn, One (Giant) Step at a Time","created":"2023-05-29","cited_by_count":68,"area":null},{"id":"2303.00055","title":"Learning time-scales in two-layers neural networks","created":"2023-02-28","cited_by_count":57,"area":null},{"id":"2201.10469","title":"Convex Analysis of the Mean Field Langevin Dynamics","created":"2022-01-25","cited_by_count":55,"area":null},{"id":"2112.15383","title":"Separation of Scales and a Thermodynamic Description of Feature Learning in Some CNNs","created":"2021-12-31","cited_by_count":53,"area":null},{"id":"2102.12781","title":"Do Input Gradients Highlight Discriminative Features?","created":"2021-02-25","cited_by_count":52,"area":null},{"id":"2310.00535","title":"JoMA: Demystifying Multilayer Transformers via JOint Dynamics of MLP and Attention","created":"2023-10-01","cited_by_count":52,"area":null},{"id":"2402.03220","title":"The Benefits of Reusing Batches for Gradient Descent in Two-Layer Networks: Breaking the Curse of Information and Leap Exponents","created":"2024-02-05","cited_by_count":52,"area":null},{"id":"2202.00293","title":"Phase diagram of Stochastic Gradient Descent in high-dimensional two-layer neural networks","created":"2022-02-01","cited_by_count":50,"area":null},{"id":"2211.11567","title":"Neural networks trained with SGD learn distributions of increasing complexity","created":"2022-11-21","cited_by_count":50,"area":null},{"id":"2405.00592","title":"Scaling and renormalization in high-dimensional regression","created":"2024-05-01","cited_by_count":50,"area":null},{"id":"2409.14623","title":"From Lazy to Rich: Exact Learning Dynamics in Deep Linear Networks","created":"2024-09-22","cited_by_count":49,"area":null},{"id":"2310.03789","title":"Grokking as a First Order Phase Transition in Two Layer Networks","created":"2023-10-05","cited_by_count":48,"area":null},{"id":"2102.02410","title":"A Local Convergence Theory for Mildly Over-Parameterized Two-Layer Neural Network","created":"2021-02-04","cited_by_count":47,"area":null},{"id":"2306.17759","title":"The Shaped Transformer: Attention Models in the Infinite Depth-and-Width Limit","created":"2023-06-30","cited_by_count":47,"area":null},{"id":"2402.01258","title":"Transformers Learn Nonlinear Features In Context: Nonconvex Mean-field Dynamics on the Attention Landscape","created":"2024-02-02","cited_by_count":46,"area":null},{"id":"2101.09612","title":"On the Proof of Global Convergence of Gradient Descent for Deep ReLU Networks with Linear Widths","created":"2021-01-24","cited_by_count":44,"area":null},{"id":"2106.03795","title":"Heavy Tails in SGD and Compressibility of Overparametrized Neural Networks","created":"2021-06-07","cited_by_count":44,"area":null},{"id":"2302.05882","title":"From high-dimensional & mean-field dynamics to dimensionless ODEs: A unifying approach to SGD in two-layers networks","created":"2023-02-12","cited_by_count":43,"area":null},{"id":"2406.06158","title":"Get rich quick: exact solutions reveal how unbalanced initializations promote rapid feature learning","created":"2024-06-10","cited_by_count":42,"area":null},{"id":"2206.02768","title":"The Neural Covariance SDE: Shaped Infinite Depth-and-Width Networks at Initialization","created":"2022-06-06","cited_by_count":41,"area":null},{"id":"2305.18411","title":"Feature-Learning Networks Are Consistent Across Widths At Realistic Scales","created":"2023-05-28","cited_by_count":41,"area":null},{"id":"2307.11353","title":"What can a Single Attention Layer Learn? A Study Through the Random Features Lens","created":"2023-07-21","cited_by_count":40,"area":null},{"id":"2309.15111","title":"SGD Finds then Tunes Features in Two-Layer Neural Networks with near-Optimal Sample Complexity: A Case Study in the XOR problem","created":"2023-09-26","cited_by_count":39,"area":null},{"id":"2410.23228","title":"Emergence of meta-stable clustering in mean-field transformer models","created":"2024-10-30","cited_by_count":39,"area":null},{"id":"2105.14573","title":"Embedding Principle of Loss Landscape of Deep Neural Networks","created":"2021-05-30","cited_by_count":38,"area":null},{"id":"2402.04980","title":"Asymptotics of feature learning in two-layer networks after one gradient-step","created":"2024-02-07","cited_by_count":38,"area":null},{"id":"2106.08929","title":"KALE Flow: A Relaxed KL Gradient Flow for Probabilities with Disjoint Support","created":"2021-06-16","cited_by_count":37,"area":null},{"id":"2207.10430","title":"The Neural Race Reduction: Dynamics of Abstraction in Gated Networks","created":"2022-07-21","cited_by_count":37,"area":null},{"id":"2306.13575","title":"Scaling MLPs: A Tale of Inductive Bias","created":"2023-06-23","cited_by_count":37,"area":null},{"id":"2405.15459","title":"Repetita Iuvant: Data Repetition Allows SGD to Learn High-Dimensional Multi-Index Functions","created":"2024-05-24","cited_by_count":37,"area":null},{"id":"2102.04396","title":"SGD in the Large: Average-case Analysis, Asymptotics, and Stepsize Criticality","created":"2021-02-08","cited_by_count":35,"area":null},{"id":"2202.07626","title":"Random Feature Amplification: Feature Learning and Generalization in Neural Networks","created":"2022-02-15","cited_by_count":35,"area":null},{"id":"2206.08873","title":"Mirror Descent with Relative Smoothness in Measure Spaces, with application to Sinkhorn and EM","created":"2022-06-17","cited_by_count":34,"area":null},{"id":"2106.04013","title":"The Future is Log-Gaussian: ResNets and Their Infinite-Depth-and-Width Limit at Initialization","created":"2021-06-07","cited_by_count":33,"area":null},{"id":"2106.07880","title":"Scaling Neural Tangent Kernels via Sketching and Random Features","created":"2021-06-15","cited_by_count":32,"area":null},{"id":"2309.03843","title":"Gradient-Based Feature Learning under Structured Data","created":"2023-09-07","cited_by_count":32,"area":null},{"id":"2111.05685","title":"Efficient Neural Network Training via Forward and Backward Propagation Sparsification","created":"2021-11-10","cited_by_count":30,"area":null},{"id":"2205.07146","title":"Trajectory Inference via Mean-field Langevin in Path Space","created":"2022-05-14","cited_by_count":30,"area":null},{"id":"2211.06506","title":"Spectral Evolution and Invariance in Linear-width Neural Networks","created":"2022-11-11","cited_by_count":30,"area":null},{"id":"2206.09841","title":"Label noise (stochastic) gradient descent implicitly solves the Lasso for quadratic parametrisation","created":"2022-06-20","cited_by_count":29,"area":null},{"id":"2604.21691","title":"There Will Be a Scientific Theory of Deep Learning","created":"2026-04-23","cited_by_count":29,"area":null},{"id":"2301.01766","title":"Learning Gaussian Mixtures Using the Wasserstein-Fisher-Rao Gradient Flow","created":"2023-01-04","cited_by_count":27,"area":null},{"id":"2105.02095","title":"Two-layer neural networks with values in a Banach space","created":"2021-05-05","cited_by_count":26,"area":null},{"id":"2206.04613","title":"Explicit Regularization in Overparametrized Models via Noise Injection","created":"2022-06-09","cited_by_count":26,"area":null},{"id":"2106.02588","title":"Stochastic gradient descent with noise of machine learning type. Part II: Continuous time analysis","created":"2021-06-04","cited_by_count":25,"area":null},{"id":"2112.10852","title":"The effective noise of Stochastic Gradient Descent","created":"2021-12-20","cited_by_count":25,"area":null},{"id":"2203.07379","title":"Quantitative Gaussian Approximation of Randomly Initialized Deep Neural Networks","created":"2022-03-14","cited_by_count":25,"area":null},{"id":"2212.12147","title":"The Onset of Variance-Limited Behavior for Networks in the Lazy and Rich Regimes","created":"2022-12-23","cited_by_count":25,"area":null},{"id":"2310.19103","title":"Proving Linear Mode Connectivity of Neural Networks via Optimal Transport","created":"2023-10-29","cited_by_count":25,"area":null},{"id":"2509.25040","title":"A multiscale analysis of mean-field transformers in the moderate interaction regime","created":"2025-09-29","cited_by_count":25,"area":null},{"id":"2105.14417","title":"Overparameterization of deep ResNet: zero loss and mean-field analysis","created":"2021-05-30","cited_by_count":24,"area":null},{"id":"2107.13289","title":"The loss landscape of deep linear neural networks: a second-order analysis","created":"2021-07-28","cited_by_count":24,"area":null},{"id":"2306.08321","title":"Nonparametric regression using over-parameterized shallow ReLU neural networks","created":"2023-06-14","cited_by_count":24,"area":null},{"id":"2310.09753","title":"When can transformers reason with abstract symbols?","created":"2023-10-15","cited_by_count":24,"area":null},{"id":"2410.19637","title":"A distributional simplicity bias in the learning dynamics of transformers","created":"2024-10-25","cited_by_count":24,"area":null},{"id":"2206.12314","title":"Learning sparse features can lead to overfitting in neural networks","created":"2022-06-24","cited_by_count":23,"area":null},{"id":"2212.03050","title":"Uniform-in-time propagation of chaos for mean field Langevin dynamics","created":"2022-12-06","cited_by_count":23,"area":null},{"id":"2306.16921","title":"Provable Advantage of Curriculum Learning on Parity Targets with Mixed Inputs","created":"2023-06-29","cited_by_count":22,"area":null},{"id":"2106.06529","title":"The Limitations of Large Width in Neural Networks: A Deep Gaussian Process Perspective","created":"2021-06-11","cited_by_count":21,"area":null},{"id":"2106.13792","title":"Proxy Convexity: A Unified Framework for the Analysis of Neural Networks Trained by Gradient Descent","created":"2021-06-25","cited_by_count":21,"area":null},{"id":"2201.07296","title":"Convergence of Policy Gradient for Entropy Regularized MDPs with Neural Network Approximation in the Mean-Field Regime","created":"2022-01-18","cited_by_count":21,"area":null},{"id":"2212.14457","title":"Bayesian Interpolation with Deep Linear Networks","created":"2022-12-29","cited_by_count":21,"area":null},{"id":"2106.06573","title":"Understanding Deflation Process in Over-parametrized Tensor Decomposition","created":"2021-06-11","cited_by_count":20,"area":null},{"id":"2109.00454","title":"The emergence of a concept in shallow neural networks","created":"2021-09-01","cited_by_count":20,"area":null},{"id":"2211.16980","title":"Infinite-width limit of deep linear neural networks","created":"2022-11-29","cited_by_count":20,"area":null},{"id":"2306.07221","title":"Convergence of mean-field Langevin dynamics: Time and space discretization, stochastic gradient, and variance reduction","created":"2023-06-12","cited_by_count":20,"area":null},{"id":"2307.06092","title":"Quantitative CLTs in Deep Neural Networks","created":"2023-07-12","cited_by_count":20,"area":null},{"id":"2308.13431","title":"Six Lectures on Linearized Neural Networks","created":"2023-08-25","cited_by_count":20,"area":null},{"id":"2401.07187","title":"A Survey on Statistical Theory of Deep Learning: Approximation, Training Dynamics, and Generative Models","created":"2024-01-14","cited_by_count":20,"area":null},{"id":"2101.02533","title":"Towards Understanding Learning in Neural Networks with Linear Teachers","created":"2021-01-07","cited_by_count":19,"area":null},{"id":"2102.08668","title":"Non-asymptotic approximations of neural networks by Gaussian processes","created":"2021-02-17","cited_by_count":19,"area":null},{"id":"2104.05508","title":"Noether: The More Things Change, the More Stay the Same","created":"2021-04-12","cited_by_count":19,"area":null},{"id":"2107.09133","title":"The Limiting Dynamics of SGD: Modified Loss, Phase Space Oscillations, and Anomalous Diffusion","created":"2021-07-19","cited_by_count":19,"area":null},{"id":"2210.02157","title":"The Influence of Learning Rule on Representation Dynamics in Wide Neural Networks","created":"2022-10-05","cited_by_count":19,"area":null}],"next_cursor":"100"}