Low-rank bias, weight decay, and model merging in neural networks

• URL: http://arxiv.org/abs/2502.17340v1
• Date: Mon, 24 Feb 2025 17:17:00 GMT
• Title: Low-rank bias, weight decay, and model merging in neural networks
• Authors: Ilja Kuzborskij, Yasin Abbasi Yadkori,
• Abstract summary: We show several properties of deep neural networks trained with $L2$ regularization.<n>We also investigate a multitask learning phenomenon enabled by $L2$ regularization and low-rank bias.
• Score: 12.352761060862072
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We explore the low-rank structure of the weight matrices in neural networks originating from training with Gradient Descent (GD) and Gradient Flow (GF) with $L2$ regularization (also known as weight decay). We show several properties of GD-trained deep neural networks, induced by $L2$ regularization. In particular, for a stationary point of GD we show alignment of the parameters and the gradient, norm preservation across layers, and low-rank bias: properties previously known in the context of GF solutions. Experiments show that the assumptions made in the analysis only mildly affect the observations. In addition, we investigate a multitask learning phenomenon enabled by $L2$ regularization and low-rank bias. In particular, we show that if two networks are trained, such that the inputs in the training set of one network are approximately orthogonal to the inputs in the training set of the other network, the new network obtained by simply summing the weights of the two networks will perform as well on both training sets as the respective individual networks. We demonstrate this for shallow ReLU neural networks trained by GD, as well as deep linear and deep ReLU networks trained by GF.

Related papers

• Implicit Bias of Gradient Descent for Two-layer ReLU and Leaky ReLU Networks on Nearly-orthogonal Data [66.1211659120882]
  The implicit bias towards solutions with favorable properties is believed to be a key reason why neural networks trained by gradient-based optimization can generalize well. While the implicit bias of gradient flow has been widely studied for homogeneous neural networks (including ReLU and leaky ReLU networks), the implicit bias of gradient descent is currently only understood for smooth neural networks.
  arXiv  Detail & Related papers  (2023-10-29T08:47:48Z)
• Adversarial Examples Exist in Two-Layer ReLU Networks for Low Dimensional Linear Subspaces [24.43191276129614]
  We show that standard methods lead to non-robust neural networks. We show that decreasing the scale of the training algorithm, or adding $L$ regularization, can make the trained network more robust to adversarial perturbations.
  arXiv  Detail & Related papers  (2023-03-01T19:10:05Z)
• Computational Complexity of Learning Neural Networks: Smoothness and Degeneracy [52.40331776572531]
  We show that learning depth-$3$ ReLU networks under the Gaussian input distribution is hard even in the smoothed-analysis framework. Our results are under a well-studied assumption on the existence of local pseudorandom generators.
  arXiv  Detail & Related papers  (2023-02-15T02:00:26Z)
• Gradient Descent in Neural Networks as Sequential Learning in RKBS [63.011641517977644]
  We construct an exact power-series representation of the neural network in a finite neighborhood of the initial weights. We prove that, regardless of width, the training sequence produced by gradient descent can be exactly replicated by regularized sequential learning.
  arXiv  Detail & Related papers  (2023-02-01T03:18:07Z)
• Neural networks trained with SGD learn distributions of increasing complexity [78.30235086565388]
  We show that neural networks trained using gradient descent initially classify their inputs using lower-order input statistics. We then exploit higher-order statistics only later during training. We discuss the relation of DSB to other simplicity biases and consider its implications for the principle of universality in learning.
  arXiv  Detail & Related papers  (2022-11-21T15:27:22Z)
• On the Effective Number of Linear Regions in Shallow Univariate ReLU Networks: Convergence Guarantees and Implicit Bias [50.84569563188485]
  We show that gradient flow converges in direction when labels are determined by the sign of a target network with $r$ neurons. Our result may already hold for mild over- parameterization, where the width is $tildemathcalO(r)$ and independent of the sample size.
  arXiv  Detail & Related papers  (2022-05-18T16:57:10Z)
• 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)
• Bayesian Nested Neural Networks for Uncertainty Calibration and Adaptive Compression [40.35734017517066]
  Nested networks or slimmable networks are neural networks whose architectures can be adjusted instantly during testing time. Recent studies have focused on a "nested dropout" layer, which is able to order the nodes of a layer by importance during training.
  arXiv  Detail & Related papers  (2021-01-27T12:34:58Z)
• A Unifying View on Implicit Bias in Training Linear Neural Networks [31.65006970108761]
  We study the implicit bias of gradient flow (i.e., gradient descent with infinitesimal step size) on linear neural network training. We propose a tensor formulation of neural networks that includes fully-connected, diagonal, and convolutional networks as special cases.
  arXiv  Detail & Related papers  (2020-10-06T06:08:35Z)

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.