Simplicity Bias of Two-Layer Networks beyond Linearly Separable Data

• URL: http://arxiv.org/abs/2405.17299v2
• Date: Thu, 07 Nov 2024 13:44:22 GMT
• Title: Simplicity Bias of Two-Layer Networks beyond Linearly Separable Data
• Authors: Nikita Tsoy, Nikola Konstantinov,
• Abstract summary: We characterize simplicity bias for general datasets in the context of two-layer neural networks with small weights and trained with gradient flow. For datasets with an XOR-like pattern, we precisely identify the learned features and demonstrate that simplicity bias intensifies during later training stages. These results indicate that features learned in the middle stages of training may be more useful for OOD transfer.
• Score: 4.14360329494344
• License:
• Abstract: Simplicity bias, the propensity of deep models to over-rely on simple features, has been identified as a potential reason for limited out-of-distribution generalization of neural networks (Shah et al., 2020). Despite the important implications, this phenomenon has been theoretically confirmed and characterized only under strong dataset assumptions, such as linear separability (Lyu et al., 2021). In this work, we characterize simplicity bias for general datasets in the context of two-layer neural networks initialized with small weights and trained with gradient flow. Specifically, we prove that in the early training phases, network features cluster around a few directions that do not depend on the size of the hidden layer. Furthermore, for datasets with an XOR-like pattern, we precisely identify the learned features and demonstrate that simplicity bias intensifies during later training stages. These results indicate that features learned in the middle stages of training may be more useful for OOD transfer. We support this hypothesis with experiments on image data.

Related papers

• Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networks [13.983863226803336]
  We argue that "Feature Averaging" is one of the principal factors contributing to non-robustness of deep neural networks. We provide a detailed theoretical analysis of the training dynamics of gradient descent in a two-layer ReLU network for a binary classification task. We prove that, with the provision of more granular supervised information, a two-layer multi-class neural network is capable of learning individual features.
  arXiv  Detail & Related papers  (2024-10-14T09:28:32Z)
• Features are fate: a theory of transfer learning in high-dimensional regression [23.840251319669907]
  We show that when the target task is well represented by the feature space of the pre-trained model, transfer learning outperforms training from scratch. For this model, we establish rigorously that when the feature space overlap between the source and target tasks is sufficiently strong, both linear transfer and fine-tuning improve performance.
  arXiv  Detail & Related papers  (2024-10-10T17:58:26Z)
• Understanding Deep Representation Learning via Layerwise Feature Compression and Discrimination [33.273226655730326]
  We show that each layer of a deep linear network progressively compresses within-class features at a geometric rate and discriminates between-class features at a linear rate. This is the first quantitative characterization of feature evolution in hierarchical representations of deep linear networks.
  arXiv  Detail & Related papers  (2023-11-06T09:00:38Z)
• 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)
• Learning an Invertible Output Mapping Can Mitigate Simplicity Bias in Neural Networks [66.76034024335833]
  We investigate why diverse/ complex features are learned by the backbone, and their brittleness is due to the linear classification head relying primarily on the simplest features. We propose Feature Reconstruction Regularizer (FRR) to ensure that the learned features can be reconstructed back from the logits. We demonstrate up to 15% gains in OOD accuracy on the recently introduced semi-synthetic datasets with extreme distribution shifts.
  arXiv  Detail & Related papers  (2022-10-04T04:01:15Z)
• Data-driven emergence of convolutional structure in neural networks [83.4920717252233]
  We show how fully-connected neural networks solving a discrimination task can learn a convolutional structure directly from their inputs. By carefully designing data models, we show that the emergence of this pattern is triggered by the non-Gaussian, higher-order local structure of the inputs.
  arXiv  Detail & Related papers  (2022-02-01T17:11:13Z)
• 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)
• On Robustness and Transferability of Convolutional Neural Networks [147.71743081671508]
  Modern deep convolutional networks (CNNs) are often criticized for not generalizing under distributional shifts. We study the interplay between out-of-distribution and transfer performance of modern image classification CNNs for the first time. We find that increasing both the training set and model sizes significantly improve the distributional shift robustness.
  arXiv  Detail & Related papers  (2020-07-16T18:39:04Z)
• Learning from Failure: Training Debiased Classifier from Biased Classifier [76.52804102765931]
  We show that neural networks learn to rely on spurious correlation only when it is "easier" to learn than the desired knowledge. We propose a failure-based debiasing scheme by training a pair of neural networks simultaneously. Our method significantly improves the training of the network against various types of biases in both synthetic and real-world datasets.
  arXiv  Detail & Related papers  (2020-07-06T07:20:29Z)
• The Surprising Simplicity of the Early-Time Learning Dynamics of Neural Networks [43.860358308049044]
  In work, we show that these common perceptions can be completely false in the early phase of learning. We argue that this surprising simplicity can persist in networks with more layers with convolutional architecture.
  arXiv  Detail & Related papers  (2020-06-25T17:42:49Z)

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.