Escaping Saddle Points for Effective Generalization on Class-Imbalanced Data

• URL: http://arxiv.org/abs/2212.13827v1
• Date: Wed, 28 Dec 2022 14:00:44 GMT
• Title: Escaping Saddle Points for Effective Generalization on Class-Imbalanced Data
• Authors: Harsh Rangwani, Sumukh K Aithal, Mayank Mishra, R. Venkatesh Babu
• Abstract summary: We analyze the class-imbalanced learning problem by examining the loss landscape of neural networks trained with re-weighting and margin-based techniques. We find that optimization methods designed to escape from saddle points can be effectively used to improve generalization on minority classes. Using SAM results in a 6.2% increase in accuracy on the minority classes over the state-of-the-art Vector Scaling Loss, leading to an overall average increase of 4% across imbalanced datasets.
• Score: 40.64419633190104
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Real-world datasets exhibit imbalances of varying types and degrees. Several techniques based on re-weighting and margin adjustment of loss are often used to enhance the performance of neural networks, particularly on minority classes. In this work, we analyze the class-imbalanced learning problem by examining the loss landscape of neural networks trained with re-weighting and margin-based techniques. Specifically, we examine the spectral density of Hessian of class-wise loss, through which we observe that the network weights converge to a saddle point in the loss landscapes of minority classes. Following this observation, we also find that optimization methods designed to escape from saddle points can be effectively used to improve generalization on minority classes. We further theoretically and empirically demonstrate that Sharpness-Aware Minimization (SAM), a recent technique that encourages convergence to a flat minima, can be effectively used to escape saddle points for minority classes. Using SAM results in a 6.2\% increase in accuracy on the minority classes over the state-of-the-art Vector Scaling Loss, leading to an overall average increase of 4\% across imbalanced datasets. The code is available at: https://github.com/val-iisc/Saddle-LongTail.

Related papers

• Class-Imbalanced Semi-Supervised Learning for Large-Scale Point Cloud Semantic Segmentation via Decoupling Optimization [64.36097398869774]
  Semi-supervised learning (SSL) has been an active research topic for large-scale 3D scene understanding. The existing SSL-based methods suffer from severe training bias due to class imbalance and long-tail distributions of the point cloud data. We introduce a new decoupling optimization framework, which disentangles feature representation learning and classifier in an alternative optimization manner to shift the bias decision boundary effectively.
  arXiv  Detail & Related papers  (2024-01-13T04:16:40Z)
• A Unified Generalization Analysis of Re-Weighting and Logit-Adjustment for Imbalanced Learning [129.63326990812234]
  We propose a technique named data-dependent contraction to capture how modified losses handle different classes. On top of this technique, a fine-grained generalization bound is established for imbalanced learning, which helps reveal the mystery of re-weighting and logit-adjustment.
  arXiv  Detail & Related papers  (2023-10-07T09:15:08Z)
• Neural Collapse Inspired Attraction-Repulsion-Balanced Loss for Imbalanced Learning [97.81549071978789]
  We propose Attraction-Repulsion-Balanced Loss (ARB-Loss) to balance the different components of the gradients. We perform experiments on the large-scale classification and segmentation datasets and our ARB-Loss can achieve state-of-the-art performance.
  arXiv  Detail & Related papers  (2022-04-19T08:23:23Z)
• Imbalanced Classification via Explicit Gradient Learning From Augmented Data [0.0]
  We propose a novel deep meta-learning technique to augment a given imbalanced dataset with new minority instances. The advantage of the proposed method is demonstrated on synthetic and real-world datasets with various imbalance ratios.
  arXiv  Detail & Related papers  (2022-02-21T22:16:50Z)
• Striking the Right Balance: Recall Loss for Semantic Segmentation [24.047359482606307]
  Class imbalance is a fundamental problem in computer vision applications such as semantic segmentation. We propose a hard-class mining loss by reshaping the vanilla cross entropy loss. We show that the novel recall loss changes gradually between the standard cross entropy loss and the inverse frequency weighted loss.
  arXiv  Detail & Related papers  (2021-06-28T18:02:03Z)
• MetaBalance: High-Performance Neural Networks for Class-Imbalanced Data [42.81296448544681]
  Class-imbalanced data, in which some classes contain far more samples than others, is ubiquitous in real-world applications. Standard techniques for handling class-imbalance usually work by training on a re-weighted loss or on re-balanced data. We evaluate our method, MetaBalance, on image classification, credit-card fraud detection, loan default prediction, and facial recognition tasks with severely imbalanced data.
  arXiv  Detail & Related papers  (2021-06-17T16:42:50Z)
• Distributional Robustness Loss for Long-tail Learning [20.800627115140465]
  Real-world data is often unbalanced and long-tailed, but deep models struggle to recognize rare classes in the presence of frequent classes. We show that the feature extractor part of deep networks suffers greatly from this bias. We propose a new loss based on robustness theory, which encourages the model to learn high-quality representations for both head and tail classes.
  arXiv  Detail & Related papers  (2021-04-07T11:34:04Z)
• Procrustean Training for Imbalanced Deep Learning [40.85940706868622]
  We show that a neural network tends to first under-fit the minor classes by classifying most of their data into the major classes. We propose a novel learning strategy to equalize the training progress across classes.
  arXiv  Detail & Related papers  (2021-04-05T04:44:01Z)
• M2m: Imbalanced Classification via Major-to-minor Translation [79.09018382489506]
  In most real-world scenarios, labeled training datasets are highly class-imbalanced, where deep neural networks suffer from generalizing to a balanced testing criterion. In this paper, we explore a novel yet simple way to alleviate this issue by augmenting less-frequent classes via translating samples from more-frequent classes. Our experimental results on a variety of class-imbalanced datasets show that the proposed method improves the generalization on minority classes significantly compared to other existing re-sampling or re-weighting methods.
  arXiv  Detail & Related papers  (2020-04-01T13:21:17Z)
• Equalization Loss for Long-Tailed Object Recognition [109.91045951333835]
  State-of-the-art object detection methods still perform poorly on large vocabulary and long-tailed datasets. We propose a simple but effective loss, named equalization loss, to tackle the problem of long-tailed rare categories. Our method achieves AP gains of 4.1% and 4.8% for the rare and common categories on the challenging LVIS benchmark.
  arXiv  Detail & Related papers  (2020-03-11T09:14:53Z)

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.