Provable Emergence of Deep Neural Collapse and Low-Rank Bias in $L^2$-Regularized Nonlinear Networks

• URL: http://arxiv.org/abs/2402.03991v2
• Date: Sat, 23 Aug 2025 20:01:08 GMT
• Title: Provable Emergence of Deep Neural Collapse and Low-Rank Bias in $L^2$-Regularized Nonlinear Networks
• Authors: Emanuele Zangrando, Piero Deidda, Simone Brugiapaglia, Nicola Guglielmi, Francesco Tudisco,
• Abstract summary: We show a link between deep neural collapse and the emergence of low-rank weight matrices for a general class of feedforward networks with nonlinear activation.<n>We also prove the global optimality of deep neural collapsed configurations and the practical absence of a loss barrier between interpolating minima and globally optimal points.
• Score: 8.271417588922336
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent work in deep learning has shown strong empirical and theoretical evidence of an implicit low-rank bias: weight matrices in deep networks tend to be approximately low-rank. Moreover, removing relatively small singular values during training, or from available trained models, may significantly reduce model size while maintaining or even improving model performance. However, the majority of the theoretical investigations around low-rank bias in neural networks deal with oversimplified models, often not taking into account the impact of nonlinearity. In this work, we first of all quantify a link between the phenomenon of deep neural collapse and the emergence of low-rank weight matrices for a general class of feedforward networks with nonlinear activation. In addition, for the general class of nonlinear feedforward and residual networks, we prove the global optimality of deep neural collapsed configurations and the practical absence of a loss barrier between interpolating minima and globally optimal points, offering a possible explanation for its common occurrence. As a byproduct, our theory also allows us to forecast the final global structure of singular values before training. Our theoretical findings are supported by a range of experimental evaluations illustrating the phenomenon.

Related papers

• Generalization performance of narrow one-hidden layer networks in the teacher-student setting [40.69556943879117]
  We develop a general theory for narrow networks, i.e. networks with a large number of hidden units, yet much smaller than the input dimension.<n>Our theory accurately predicts the generalization error of neural networks trained on regression or classification tasks.
  arXiv  Detail & Related papers  (2025-07-01T10:18:20Z)
• Towards the Training of Deeper Predictive Coding Neural Networks [53.15874572081944]
  Predictive coding networks trained with equilibrium propagation are neural models that perform inference through an iterative energy process.<n>Previous studies have demonstrated their effectiveness in shallow architectures, but show significant performance degradation when depth exceeds five to seven layers.<n>We show that the reason behind this degradation is due to exponentially imbalanced errors between layers during weight updates, and predictions from the previous layer not being effective in guiding updates in deeper layers.
  arXiv  Detail & Related papers  (2025-06-30T12:44:47Z)
• An Overview of Low-Rank Structures in the Training and Adaptation of Large Models [52.67110072923365]
  Recent research has uncovered a widespread phenomenon in deep networks: the emergence of low-rank structures. These implicit low-dimensional patterns provide valuable insights for improving the efficiency of training and fine-tuning large-scale models. We present a comprehensive review of advances in exploiting low-rank structures for deep learning and shed light on their mathematical foundations.
  arXiv  Detail & Related papers  (2025-03-25T17:26:09Z)
• Low-rank bias, weight decay, and model merging in neural networks [12.352761060862072]
  We show several properties of deep neural networks trained with $L2$ regularization. We also investigate a multitask learning phenomenon enabled by $L2$ regularization and low-rank bias.
  arXiv  Detail & Related papers  (2025-02-24T17:17:00Z)
• On the Dynamics Under the Unhinged Loss and Beyond [104.49565602940699]
  We introduce the unhinged loss, a concise loss function, that offers more mathematical opportunities to analyze closed-form dynamics. The unhinged loss allows for considering more practical techniques, such as time-vary learning rates and feature normalization.
  arXiv  Detail & Related papers  (2023-12-13T02:11:07Z)
• Addressing caveats of neural persistence with deep graph persistence [54.424983583720675]
  We find that the variance of network weights and spatial concentration of large weights are the main factors that impact neural persistence. We propose an extension of the filtration underlying neural persistence to the whole neural network instead of single layers. This yields our deep graph persistence measure, which implicitly incorporates persistent paths through the network and alleviates variance-related issues.
  arXiv  Detail & Related papers  (2023-07-20T13:34:11Z)
• Sparsity-aware generalization theory for deep neural networks [12.525959293825318]
  We present a new approach to analyzing generalization for deep feed-forward ReLU networks. We show fundamental trade-offs between sparsity and generalization.
  arXiv  Detail & Related papers  (2023-07-01T20:59:05Z)
• Rank Diminishing in Deep Neural Networks [71.03777954670323]
  Rank of neural networks measures information flowing across layers. It is an instance of a key structural condition that applies across broad domains of machine learning. For neural networks, however, the intrinsic mechanism that yields low-rank structures remains vague and unclear.
  arXiv  Detail & Related papers  (2022-06-13T12:03:32Z)
• SGD and Weight Decay Secretly Minimize the Rank of Your Neural Network [8.79431718760617]
  Training with mini-batch SGD and weight decay induces a bias toward rank minimization in weight matrices. We show that this bias becomes more pronounced with smaller batch sizes, higher learning rates, or stronger weight decay. We empirically explore the connection between this bias and generalization, finding that it has a marginal effect on the test performance.
  arXiv  Detail & Related papers  (2022-06-12T17:06:35Z)
• With Greater Distance Comes Worse Performance: On the Perspective of Layer Utilization and Model Generalization [3.6321778403619285]
  Generalization of deep neural networks remains one of the main open problems in machine learning. Early layers generally learn representations relevant to performance on both training data and testing data. Deeper layers only minimize training risks and fail to generalize well with testing or mislabeled data.
  arXiv  Detail & Related papers  (2022-01-28T05:26:32Z)
• Multi-scale Feature Learning Dynamics: Insights for Double Descent [71.91871020059857]
  We study the phenomenon of "double descent" of the generalization error. We find that double descent can be attributed to distinct features being learned at different scales.
  arXiv  Detail & Related papers  (2021-12-06T18:17:08Z)
• 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)
• Compact representations of convolutional neural networks via weight pruning and quantization [63.417651529192014]
  We propose a novel storage format for convolutional neural networks (CNNs) based on source coding and leveraging both weight pruning and quantization. We achieve a reduction of space occupancy up to 0.6% on fully connected layers and 5.44% on the whole network, while performing at least as competitive as the baseline.
  arXiv  Detail & Related papers  (2021-08-28T20:39:54Z)
• The Interplay Between Implicit Bias and Benign Overfitting in Two-Layer Linear Networks [51.1848572349154]
  neural network models that perfectly fit noisy data can generalize well to unseen test data. We consider interpolating two-layer linear neural networks trained with gradient flow on the squared loss and derive bounds on the excess risk.
  arXiv  Detail & Related papers  (2021-08-25T22:01:01Z)
• Learning and Generalization in Overparameterized Normalizing Flows [13.074242275886977]
  Normalizing flows (NFs) constitute an important class of models in unsupervised learning. We provide theoretical and empirical evidence that for a class of NFs containing most of the existing NF models, overparametrization hurts training. We prove that unconstrained NFs can efficiently learn any reasonable data distribution under minimal assumptions when the underlying network is overparametrized.
  arXiv  Detail & Related papers  (2021-06-19T17:11:42Z)
• Gradient-trained Weights in Wide Neural Networks Align Layerwise to Error-scaled Input Correlations [11.176824373696324]
  We derive the layerwise weight dynamics of infinite-width neural networks with nonlinear activations trained by gradient descent. We formulate backpropagation-free learning rules, named Align-zero and Align-ada, that theoretically achieve the same alignment as backpropagation.
  arXiv  Detail & Related papers  (2021-06-15T21:56:38Z)
• Predicting Unreliable Predictions by Shattering a Neural Network [145.3823991041987]
  Piecewise linear neural networks can be split into subfunctions. Subfunctions have their own activation pattern, domain, and empirical error. Empirical error for the full network can be written as an expectation over subfunctions.
  arXiv  Detail & Related papers  (2021-06-15T18:34:41Z)
• The Low-Rank Simplicity Bias in Deep Networks [46.79964271742486]
  We make a series of empirical observations that investigate and extend the hypothesis that deep networks are inductively biased to find solutions with lower effective rank embeddings. We show that our claim holds true on finite width linear and non-linear models on practical learning paradigms and show that on natural data, these are often the solutions that generalize well.
  arXiv  Detail & Related papers  (2021-03-18T17:58:02Z)
• Formalizing Generalization and Robustness of Neural Networks to Weight Perturbations [58.731070632586594]
  We provide the first formal analysis for feed-forward neural networks with non-negative monotone activation functions against weight perturbations. We also design a new theory-driven loss function for training generalizable and robust neural networks against weight perturbations.
  arXiv  Detail & Related papers  (2021-03-03T06:17:03Z)
• Exploring Deep Neural Networks via Layer-Peeled Model: Minority Collapse in Imbalanced Training [39.137793683411424]
  We introduce the textitLayer-Peeled Model, a non-yet analytically tractable optimization program. We show that the model inherits many characteristics of well-trained networks, thereby offering an effective tool for explaining and predicting common empirical patterns of deep learning training. In particular, we show that the model reveals a hitherto unknown phenomenon that we term textitMinority Collapse, which fundamentally limits the performance of deep learning models on the minority classes.
  arXiv  Detail & Related papers  (2021-01-29T17:37:17Z)
• Gradient Starvation: A Learning Proclivity in Neural Networks [97.02382916372594]
  Gradient Starvation arises when cross-entropy loss is minimized by capturing only a subset of features relevant for the task. This work provides a theoretical explanation for the emergence of such feature imbalance in neural networks.
  arXiv  Detail & Related papers  (2020-11-18T18:52:08Z)

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.