Should We Always Train Models on Fine-Grained Classes?

• URL: http://arxiv.org/abs/2509.05130v1
• Date: Fri, 05 Sep 2025 14:15:46 GMT
• Title: Should We Always Train Models on Fine-Grained Classes?
• Authors: Davide Pirovano, Federico Milanesio, Michele Caselle, Piero Fariselli, Matteo Osella,
• Abstract summary: We show that training on fine-grained labels does not universally improve classification accuracy.<n>The effectiveness of this strategy depends critically on the geometric structure of the data and its relations with the label hierarchy.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In classification problems, models must predict a class label based on the input data features. However, class labels are organized hierarchically in many datasets. While a classification task is often defined at a specific level of this hierarchy, training can utilize a finer granularity of labels. Empirical evidence suggests that such fine-grained training can enhance performance. In this work, we investigate the generality of this observation and explore its underlying causes using both real and synthetic datasets. We show that training on fine-grained labels does not universally improve classification accuracy. Instead, the effectiveness of this strategy depends critically on the geometric structure of the data and its relations with the label hierarchy. Additionally, factors such as dataset size and model capacity significantly influence whether fine-grained labels provide a performance benefit.

Related papers

• Posterior Label Smoothing for Node Classification [2.737276507021477]
  We propose a simple yet effective label smoothing for the transductive node classification task. We design the soft label to encapsulate the local context of the target node through the neighborhood label distribution. In the following analysis, we find that incorporating global label statistics in posterior computation is the key to the success of label smoothing.
  arXiv  Detail & Related papers  (2024-06-01T11:59:49Z)
• Association Graph Learning for Multi-Task Classification with Category Shifts [68.58829338426712]
  We focus on multi-task classification, where related classification tasks share the same label space and are learned simultaneously. We learn an association graph to transfer knowledge among tasks for missing classes. Our method consistently performs better than representative baselines.
  arXiv  Detail & Related papers  (2022-10-10T12:37:41Z)
• Preserving Fine-Grain Feature Information in Classification via Entropic Regularization [10.358087436626391]
  We show that standard cross-entropy can lead to overfitting to coarse-related features. We introduce an entropy-based regularization to promote more diversity in the feature space of trained models.
  arXiv  Detail & Related papers  (2022-08-07T09:25:57Z)
• Use All The Labels: A Hierarchical Multi-Label Contrastive Learning Framework [75.79736930414715]
  We present a hierarchical multi-label representation learning framework that can leverage all available labels and preserve the hierarchical relationship between classes. We introduce novel hierarchy preserving losses, which jointly apply a hierarchical penalty to the contrastive loss, and enforce the hierarchy constraint.
  arXiv  Detail & Related papers  (2022-04-27T21:41:44Z)
• Debiased Pseudo Labeling in Self-Training [77.83549261035277]
  Deep neural networks achieve remarkable performances on a wide range of tasks with the aid of large-scale labeled datasets. To mitigate the requirement for labeled data, self-training is widely used in both academia and industry by pseudo labeling on readily-available unlabeled data. We propose Debiased, in which the generation and utilization of pseudo labels are decoupled by two independent heads.
  arXiv  Detail & Related papers  (2022-02-15T02:14:33Z)
• Highly Efficient Representation and Active Learning Framework for Imbalanced Data and its Application to COVID-19 X-Ray Classification [0.7829352305480284]
  We propose a highly data-efficient classification and active learning framework for classifying chest X-rays. It is based on (1) unsupervised representation learning of a Convolutional Neural Network and (2) the Gaussian Process method. We demonstrate that only $sim 10%$ of the labeled data is needed to reach the accuracy from training all available labels.
  arXiv  Detail & Related papers  (2021-02-25T02:48:59Z)
• Inducing a hierarchy for multi-class classification problems [11.58041597483471]
  In applications where categorical labels follow a natural hierarchy, classification methods that exploit the label structure often outperform those that do not. In this paper, we investigate a class of methods that induce a hierarchy that can similarly improve classification performance over flat classifiers. We demonstrate the effectiveness of the class of methods both for discovering a latent hierarchy and for improving accuracy in principled simulation settings and three real data applications.
  arXiv  Detail & Related papers  (2021-02-20T05:40:42Z)
• Label Confusion Learning to Enhance Text Classification Models [3.0251266104313643]
  Label Confusion Model (LCM) learns label confusion to capture semantic overlap among labels. LCM can generate a better label distribution to replace the original one-hot label vector. experiments on five text classification benchmark datasets reveal the effectiveness of LCM for several widely used deep learning classification models.
  arXiv  Detail & Related papers  (2020-12-09T11:34:35Z)
• Text Classification Using Label Names Only: A Language Model Self-Training Approach [80.63885282358204]
  Current text classification methods typically require a good number of human-labeled documents as training data. We show that our model achieves around 90% accuracy on four benchmark datasets including topic and sentiment classification.
  arXiv  Detail & Related papers  (2020-10-14T17:06:41Z)
• Hierarchical Image Classification using Entailment Cone Embeddings [68.82490011036263]
  We first inject label-hierarchy knowledge into an arbitrary CNN-based classifier. We empirically show that availability of such external semantic information in conjunction with the visual semantics from images boosts overall performance.
  arXiv  Detail & Related papers  (2020-04-02T10:22:02Z)
• Structured Prediction with Partial Labelling through the Infimum Loss [85.4940853372503]
  The goal of weak supervision is to enable models to learn using only forms of labelling which are cheaper to collect. This is a type of incomplete annotation where, for each datapoint, supervision is cast as a set of labels containing the real one. This paper provides a unified framework based on structured prediction and on the concept of infimum loss to deal with partial labelling.
  arXiv  Detail & Related papers  (2020-03-02T13:59:41Z)

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.