Implicit Bias of SGD in Multivariate ReLU Networks: Effective Width Collapse

• URL: http://arxiv.org/abs/2607.03613v1
• Date: Fri, 03 Jul 2026 22:03:13 GMT
• Title: Implicit Bias of SGD in Multivariate ReLU Networks: Effective Width Collapse
• Abstract summary: We show that the input weights and biases align along finitely many directions, leading to an effective width collapse.<n>We establish a non-redundancy property of the learned representation by proving that each learned direction induces a unique ternary activation pattern on the training data.
• Score: 16.78534503577613
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We study the implicit bias of noisy stochastic gradient descent in training wide two-layer ReLU networks for multivariate regression. In a mean-field regime, the training dynamics are approximated by a Wasserstein gradient flow that converges to a unique stationary measure. We characterize the structure of this stationary measure and the predictor it represents. We show that, despite the network being infinitely overparameterized, the learned predictor admits an effectively finite representation: the input weights and biases align along finitely many directions, leading to an effective width collapse. In particular, the solution function is continuous piecewise affine, with affine regions determined by the cells of a finite hyperplane arrangement. The number of learned directions, and hence hyperplanes, is bounded above by $2\mathcal{P}-1$, where $\mathcal{P}$ denotes the number of linear dichotomies realizable on the training inputs. We further establish a non-redundancy property of the learned representation by proving that each learned direction induces a unique ternary activation pattern on the training data. Consequently, the complexity of the learned predictor is governed by the combinatorial geometry of the training data.

Related papers

• On Interpolation Formulas Describing Neural Network Generalization [8.154667543705687]
  We introduce a gradient kernel that extends the deterministic version via a continuous-time diffusion.<n>It reveals that training samples contribute through loss-dependent weights and gradient alignment along the trajectory.<n>We visualize then link the generalization error to the null space of the integral operator induced by the gradient kernel.
  arXiv  Detail & Related papers  (2026-03-14T10:03:29Z)
• Feature learning in finite-width Bayesian deep linear networks with multiple outputs and convolutional layers [39.71511919246829]
  Deep linear networks have been extensively studied, but little is known in the case of finite-width architectures with multiple outputs and convolutional layers.<n>Our work provides a dictionary that translates this physics intuition and terminology into rigorous Bayesian statistics.
  arXiv  Detail & Related papers  (2024-06-05T13:37:42Z)
• Adaptive Federated Learning Over the Air [108.62635460744109]
  We propose a federated version of adaptive gradient methods, particularly AdaGrad and Adam, within the framework of over-the-air model training. Our analysis shows that the AdaGrad-based training algorithm converges to a stationary point at the rate of $mathcalO( ln(T) / T 1 - frac1alpha ).
  arXiv  Detail & Related papers  (2024-03-11T09:10:37Z)
• The Implicit Bias of Minima Stability in Multivariate Shallow ReLU Networks [53.95175206863992]
  We study the type of solutions to which gradient descent converges when used to train a single hidden-layer multivariate ReLU network with the quadratic loss. We prove that although shallow ReLU networks are universal approximators, stable shallow networks are not.
  arXiv  Detail & Related papers  (2023-06-30T09:17:39Z)
• Training invariances and the low-rank phenomenon: beyond linear networks [44.02161831977037]
  We show that when one trains a deep linear network with logistic or exponential loss on linearly separable data, the weights converge to rank-$1$ matrices. This is the first time a low-rank phenomenon is proven rigorously for nonlinear ReLU-activated feedforward networks. Our proof relies on a specific decomposition of the network into a multilinear function and another ReLU network whose weights are constant under a certain parameter directional convergence.
  arXiv  Detail & Related papers  (2022-01-28T07:31:19Z)
• Mean-field Analysis of Piecewise Linear Solutions for Wide ReLU Networks [83.58049517083138]
  We consider a two-layer ReLU network trained via gradient descent. We show that SGD is biased towards a simple solution. We also provide empirical evidence that knots at locations distinct from the data points might occur.
  arXiv  Detail & Related papers  (2021-11-03T15:14:20Z)
• Deep Learning Approximation of Diffeomorphisms via Linear-Control Systems [91.3755431537592]
  We consider a control system of the form $dot x = sum_i=1lF_i(x)u_i$, with linear dependence in the controls. We use the corresponding flow to approximate the action of a diffeomorphism on a compact ensemble of points.
  arXiv  Detail & Related papers  (2021-10-24T08:57:46Z)
• Implicit Bias of Gradient Descent for Mean Squared Error Regression with Two-Layer Wide Neural Networks [1.3706331473063877]
  We show that the solution of training a width-$n$ shallow ReLU network is within $n- 1/2$ of the function which fits the training data. We also show that the training trajectories are captured by trajectories of smoothing splines with decreasing regularization strength.
  arXiv  Detail & Related papers  (2020-06-12T17:46:40Z)
• Path Sample-Analytic Gradient Estimators for Stochastic Binary Networks [78.76880041670904]
  In neural networks with binary activations and or binary weights the training by gradient descent is complicated. We propose a new method for this estimation problem combining sampling and analytic approximation steps. We experimentally show higher accuracy in gradient estimation and demonstrate a more stable and better performing training in deep convolutional models.
  arXiv  Detail & Related papers  (2020-06-04T21:51:21Z)
• Semiparametric Nonlinear Bipartite Graph Representation Learning with Provable Guarantees [106.91654068632882]
  We consider the bipartite graph and formalize its representation learning problem as a statistical estimation problem of parameters in a semiparametric exponential family distribution. We show that the proposed objective is strongly convex in a neighborhood around the ground truth, so that a gradient descent-based method achieves linear convergence rate. Our estimator is robust to any model misspecification within the exponential family, which is validated in extensive experiments.
  arXiv  Detail & Related papers  (2020-03-02T16:40:36Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.