Hierarchical Classification by Training to Diffuse on the Manifold

Kaixiang Song, Guoyuan An, Yuchi Huo, Yanan Li, Wei Hua, Sung-eui Yoon, Shu KongOpenReview 2024 · Submission to ICLR 2024 · Withdrawn · OpenReview

Its authors withdrew this paper, so there is no full text to read or summarize.

Abstract

Hierarchical classification, the problem of requiring classifying images according to a hierarchical taxonomy, has broad applications owing to the principle of ``making better mistakes'', i.e., better to predict correct coarse labels than incorrect fine labels. Despite the importance, the literature has found it sufficient to use the wide-adopted top-1 classification accuracy to rank methods and unnecessary to use hierarchical metrics. Importantly, the method of training deep neural networks using CE loss (or a flat softmax classifier on leaf classes only) performs well, and other ad-hoc methods do not necessarily rival the flat-softmax method. As a result, hierarchical classification has been under-explored and there lacks training and inference methods for this problem. In this paper, we study hierarchical classification from a novel perspective of hierarchical manifolds, assuming data from a hierarchical taxonomy lie in a hierarchical manifold in the feature space. This motivates our novel strategies for training models of hierarchical clasification and inference. For training, we propose a hierarchical cross-modal contrastive learning method to finetune a vision-language pre-trained model, aiming to craft an embedding space that mirrors hierarchical taxonomy. For inference, we present a diffusion method, re-conceptualizing hierarchical classification by treating its structure as a graph. Our method distinguishes from the conventional top-down and bottom-up methods for hierarchical classification. Extensive experiments validate our methods, achieving the state-of-the-art on two large-scale datasets, iNaturalist2018 and iNaturalist2021, with respect to both the top-1 accuracy and diverse hierarchical metrics.
BibTeX
@misc{song2024hierarchical,
  author = {Kaixiang Song and Guoyuan An and Yuchi Huo and Yanan Li and Wei Hua and Sung-eui Yoon and Shu Kong},
  title = {{Hierarchical Classification by Training to Diffuse on the Manifold}},
  year = {2024},
  url = {https://openreview.net/forum?id=RUgBoMu0ad},
  howpublished = {OpenReview},
  note = {Submitted to ICLR 2024},
}
Figure