On the global convergence of gradient descent for wide shallow models with bounded nonlinearities

• URL: http://arxiv.org/abs/2605.10775v1
• Date: Mon, 11 May 2026 16:08:23 GMT
• Title: On the global convergence of gradient descent for wide shallow models with bounded nonlinearities
• Abstract summary: A surprising phenomenon in neural networks is the ability of gradient descent to find global minimizers of the training loss.<n>We show that all non-global minimizers of the training loss are unstable gradient descent dynamics.
• Score: 18.375053625992802
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: A surprising phenomenon in the training of neural networks is the ability of gradient descent to find global minimizers of the training loss despite its non-convexity. Following earlier works, we investigate this behavior for wide shallow networks. Existing results essentially cover the case of ReLU activations and the case of sigmoid activations with scalar output weights. We study a large class of models that includes multi-head attention layers and two-layer sigmoid networks with vector output weights. Building upon [Chizat and Bach, 2018], we prove that all non-global minimizers of the training loss are unstable under gradient descent dynamics. Thus, when the initial distribution of the parameters has full support (which includes the popular Gaussian case), and in the many hidden neurons or attention heads limit, continuous-time gradient descent can only converge to global minimizers. Establishing the instability of non-global minimizers corresponds to the construction of an ``escaping active set'' -- we complete the proof of [Chizat and Bach, 2018] to construct this set for models with bounded nonlinearities and scalar output weights. We also extend this construction to new cases for models with vector output weights. Finally, we show the well-posedness and the stability with respect to discretization of the mean field training dynamic for sub-Gaussian initializations.

Related papers

• Across the Loss Landscape with Progressive Growth [51.366966420881646]
  We show how grow-and-optimize strategies to train toward flatter regions by viewing growth as progressive relaxation.<n>We empirically validate these predictions in controlled toy basins and in a realistic ResNet/CI-100 setting.<n>Although progressive subspace growth reliably produces flatter solutions, curvature reductions do not universally translate into improved test performance, highlighting subtleties in the flatness-generalization connection.
  arXiv  Detail & Related papers  (2026-08-25T13:50:48Z)
• When Both Layers Learn: Training Dynamics of Representing Linear Models via ReLU Networks [22.81761236732655]
  We study the gradient descent dynamics for jointly training both layers of a one-hidden-layer ReLU network to fit a linear target function.<n>Our analysis tracks the trajectory through three phases: an alignment phase in which hidden weights progressively align with the planted direction while the output weights maintain the correct sign pattern.<n>We establish novel uniform concentration results that hold along the entire trajectory, and are essential for obtaining order-wise optimal sample complexity.
  arXiv  Detail & Related papers  (2026-06-03T05:44:30Z)
• Non-Singularity of the Gradient Descent map for Neural Networks with Piecewise Analytic Activations [53.348574336527854]
  We investigate the neural network map as a function on the space of weights and biases.<n>We prove, for the first time, the non-singularity of the gradient descent (GD) map on the loss landscape of realistic neural network architectures.
  arXiv  Detail & Related papers  (2025-10-28T14:34:33Z)
• Description of the Training Process of Neural Networks via Ergodic Theorem : Ghost nodes [3.637162892228131]
  We present a unified framework for understanding and accelerating deep neural networks via training gradient descent (SGD)<n>We introduce a practical diagnostic, the running estimate of the largest Lyapunov exponent, which distinguishes genuine convergence toward stablers.<n>We propose a ghost category extension for standard classifiers that adds auxiliary ghost output nodes so the model gains extra descent directions.
  arXiv  Detail & Related papers  (2025-07-01T17:54:35Z)
• The global convergence time of stochastic gradient descent in non-convex landscapes: Sharp estimates via large deviations [29.642830843568525]
  We study the time it takes for the gradient of descent to reach the global minimum of a general, non- loss function.<n>Motivated by applications to neural networks, we provide a series of refinements and extensions of our analysis for loss functions with local minima.
  arXiv  Detail & Related papers  (2025-03-20T17:54:04Z)
• On the Convergence of Gradient Descent for Large Learning Rates [55.33626480243135]
  We show that convergence is impossible when a fixed step size is used.<n>We provide a proof of this in the case of linear neural networks with a squared loss.<n>We also prove the impossibility of convergence for more general losses without requiring strong assumptions such as Lipschitz continuity for the gradient.
  arXiv  Detail & Related papers  (2024-02-20T16:01:42Z)
• Convergence of mean-field Langevin dynamics: Time and space discretization, stochastic gradient, and variance reduction [49.66486092259376]
  The mean-field Langevin dynamics (MFLD) is a nonlinear generalization of the Langevin dynamics that incorporates a distribution-dependent drift. Recent works have shown that MFLD globally minimizes an entropy-regularized convex functional in the space of measures. We provide a framework to prove a uniform-in-time propagation of chaos for MFLD that takes into account the errors due to finite-particle approximation, time-discretization, and gradient approximation.
  arXiv  Detail & Related papers  (2023-06-12T16:28:11Z)
• Perturbation Analysis of Neural Collapse [24.94449183555951]
  Training deep neural networks for classification often includes minimizing the training loss beyond the zero training error point. Recent works analyze this behavior via idealized unconstrained features models where all the minimizers exhibit exact collapse. We propose a richer model that can capture this phenomenon by forcing the features to stay in the vicinity of a predefined features matrix.
  arXiv  Detail & Related papers  (2022-10-29T17:46:03Z)
• On Feature Learning in Neural Networks with Global Convergence Guarantees [49.870593940818715]
  We study the optimization of wide neural networks (NNs) via gradient flow (GF) We show that when the input dimension is no less than the size of the training set, the training loss converges to zero at a linear rate under GF. We also show empirically that, unlike in the Neural Tangent Kernel (NTK) regime, our multi-layer model exhibits feature learning and can achieve better generalization performance than its NTK counterpart.
  arXiv  Detail & Related papers  (2022-04-22T15:56:43Z)
• Extended Unconstrained Features Model for Exploring Deep Neural Collapse [59.59039125375527]
  Recently, a phenomenon termed "neural collapse" (NC) has been empirically observed in deep neural networks. Recent papers have shown that minimizers with this structure emerge when optimizing a simplified "unconstrained features model" In this paper, we study the UFM for the regularized MSE loss, and show that the minimizers' features can be more structured than in the cross-entropy case.
  arXiv  Detail & Related papers  (2022-02-16T14:17:37Z)
• An Unconstrained Layer-Peeled Perspective on Neural Collapse [20.75423143311858]
  We introduce a surrogate model called the unconstrained layer-peeled model (ULPM) We prove that gradient flow on this model converges to critical points of a minimum-norm separation problem exhibiting neural collapse in its global minimizer. We show that our results also hold during the training of neural networks in real-world tasks when explicit regularization or weight decay is not used.
  arXiv  Detail & Related papers  (2021-10-06T14:18:47Z)
• Kernel and Rich Regimes in Overparametrized Models [69.40899443842443]
  We show that gradient descent on overparametrized multilayer networks can induce rich implicit biases that are not RKHS norms. We also demonstrate this transition empirically for more complex matrix factorization models and multilayer non-linear networks.
  arXiv  Detail & Related papers  (2020-02-20T15:43:02Z)

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.