Rethinking Bregman Divergences in Kronecker-Factored Optimizers

• URL: http://arxiv.org/abs/2606.00542v2
• Date: Tue, 02 Jun 2026 15:25:25 GMT
• Title: Rethinking Bregman Divergences in Kronecker-Factored Optimizers
• Abstract summary: We show that Frobenius, von Neumann, and LogDet divergences distribute the unavoidable Kronecker approximation error differently.<n>We propose a subspace-aware Kronecker that applies eigenvalue-based preconditioning in the top subspace and uses an adaptive isotropic acceleration constant in the bottom subspace.
• Score: 9.141684114595433
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Shampoo-style optimizers approximate gradient covariance matrices using Kronecker-factored structures. Recent work~\cite{lin2026understanding} showed that such approximations can be viewed as projections under Bregman matrix divergences, leading to different Kronecker-factored preconditioners. However, it remains unclear what role the choice of divergence plays when the covariance is not exactly Kronecker-factored. We study this question through the spectrum of the covariance matrix. We show that Frobenius, von Neumann, and LogDet divergences distribute the unavoidable Kronecker approximation error differently across the covariance spectrum. We further show that their Kronecker factors are governed by divergence-weighted residuals rather than the raw approximation error, explaining how these spectral preferences are realized in the resulting preconditioners. Empirically, we observe that the top covariance eigenspace is substantially better aligned with the Hessian matrix, while the tail spectrum is much noisier and unreliable. Motivated by these findings, we propose a subspace-aware Kronecker optimizer that applies eigenvalue-based preconditioning in the top subspace and uses an adaptive isotropic acceleration constant in the bottom subspace.

Related papers

• Inference and Uncertainty Quantification for Streaming $r$-PCA [0.40105987447353786]
  We address two open questions in streaming PCA via Oja's algorithm.<n>We prove a matching lower bound, up to logarithmic factors, across both dense-tail sparse-tail regimes.
  arXiv  Detail & Related papers  (2026-08-18T23:02:28Z)
• The Zero Pattern of a Design Matrix Drives Multiple Descent in Over-parameterized Regression [4.064516150252115]
  We show that degeneracy of the covariance matrices and dependence can lead to multiple descent.<n>We show that maximum matching proofs and the Dulmage--Mendelsohn decomposition of the associated bipartite graph identify the configurations at which the variance becomes singular.
  arXiv  Detail & Related papers  (2026-07-27T06:30:49Z)
• Row-stochastic matrices can provably outperform doubly stochastic matrices in decentralized learning [10.686669655748702]
  Decentralized learning often involves a weighted global loss with heterogeneous node weights $$.<n>We develop a weighted Hilbert-space framework $L2(mathbbRd)$ and obtain convergence rates that are strictly tighter than those from Euclidean analysis.<n>We then derive sufficient conditions under which the row-stochastic design converges faster even with a smaller spectral gap.
  arXiv  Detail & Related papers  (2025-11-24T02:58:38Z)
• Theoretical Guarantees for Variational Inference with Fixed-Variance Mixture of Gaussians [27.20127082606962]
  Variational inference (VI) is a popular approach in Bayesian inference. This work aims to contribute to the theoretical study of VI in the non-Gaussian case.
  arXiv  Detail & Related papers  (2024-06-06T12:38:59Z)
• A Geometric Unification of Distributionally Robust Covariance Estimators: Shrinking the Spectrum by Inflating the Ambiguity Set [20.166217494056916]
  We propose a principled approach to construct covariance estimators without imposing restrictive assumptions.<n>We show that our robust estimators are efficiently computable and consistent.<n> Numerical experiments based on synthetic and real data show that our robust estimators are competitive with state-of-the-art estimators.
  arXiv  Detail & Related papers  (2024-05-30T15:01:18Z)
• Variance-Reducing Couplings for Random Features [57.73648780299374]
  Random features (RFs) are a popular technique to scale up kernel methods in machine learning. We find couplings to improve RFs defined on both Euclidean and discrete input spaces. We reach surprising conclusions about the benefits and limitations of variance reduction as a paradigm.
  arXiv  Detail & Related papers  (2024-05-26T12:25:09Z)
• Entrywise error bounds for low-rank approximations of kernel matrices [55.524284152242096]
  We derive entrywise error bounds for low-rank approximations of kernel matrices obtained using the truncated eigen-decomposition. A key technical innovation is a delocalisation result for the eigenvectors of the kernel matrix corresponding to small eigenvalues. We validate our theory with an empirical study of a collection of synthetic and real-world datasets.
  arXiv  Detail & Related papers  (2024-05-23T12:26:25Z)
• Intrinsic Bayesian Cramér-Rao Bound with an Application to Covariance Matrix Estimation [49.67011673289242]
  This paper presents a new performance bound for estimation problems where the parameter to estimate lies in a smooth manifold. It induces a geometry for the parameter manifold, as well as an intrinsic notion of the estimation error measure.
  arXiv  Detail & Related papers  (2023-11-08T15:17:13Z)
• Spectrum-Aware Debiasing: A Modern Inference Framework with Applications to Principal Components Regression [1.342834401139078]
  We introduce SpectrumAware Debiasing, a novel method for high-dimensional regression. Our approach applies to problems with structured, heavy tails, and low-rank structures. We demonstrate our method through simulated and real data experiments.
  arXiv  Detail & Related papers  (2023-09-14T15:58:30Z)
• Equivariance Discovery by Learned Parameter-Sharing [153.41877129746223]
  We study how to discover interpretable equivariances from data. Specifically, we formulate this discovery process as an optimization problem over a model's parameter-sharing schemes. Also, we theoretically analyze the method for Gaussian data and provide a bound on the mean squared gap between the studied discovery scheme and the oracle scheme.
  arXiv  Detail & Related papers  (2022-04-07T17:59:19Z)
• Optimal policy evaluation using kernel-based temporal difference methods [78.83926562536791]
  We use kernel Hilbert spaces for estimating the value function of an infinite-horizon discounted Markov reward process. We derive a non-asymptotic upper bound on the error with explicit dependence on the eigenvalues of the associated kernel operator. We prove minimax lower bounds over sub-classes of MRPs.
  arXiv  Detail & Related papers  (2021-09-24T14:48:20Z)
• Spectral clustering under degree heterogeneity: a case for the random walk Laplacian [83.79286663107845]
  This paper shows that graph spectral embedding using the random walk Laplacian produces vector representations which are completely corrected for node degree. In the special case of a degree-corrected block model, the embedding concentrates about K distinct points, representing communities.
  arXiv  Detail & Related papers  (2021-05-03T16:36:27Z)

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.