AI Heap
Published on

Disentanglement in Difference: Directly Learning Semantically Disentangled Representations by Maximizing Inter-Factor Differences

arXiv:2502.03123 - [arXiv,PDF]
Authors
  • Name
    Xingshen Zhang
  • Name
    Lin Wang
  • Name
    Shuangrong Liu
  • Name
    Xintao Lu
  • Name
    Chaoran Pang
  • Name
    Bo Yang
  • Affiliation
    Department of Computer Science, University of XYZ
  • Affiliation
    Department of Mathematics, University of ABC
  • Affiliation
    Department of Physics, University of DEF
  • Affiliation
    Department of Engineering, University of GHI
  • Affiliation
    Department of Biology, University of JKL
  • Affiliation
    Department of Chemistry, University of MNO
In this study, Disentanglement in Difference(DiD) is proposed to address the inherent inconsistency between the statistical independence of latent variables and the goal of semantic disentanglement in disentanglement representation learning. Conventional disentanglement methods achieve disentanglement representation by improving statistical independence among latent variables. However, the statistical independence of latent variables does not necessarily imply that they are semantically unrelated, thus, improving statistical independence does not always enhance disentanglement performance. To address the above issue, DiD is proposed to directly learn semantic differences rather than the statistical independence of latent variables. In the DiD, a Difference Encoder is designed to measure the semantic differences; a contrastive loss function is established to facilitate inter-dimensional comparison. Both of them allow the model to directly differentiate and disentangle distinct semantic factors, thereby resolving the inconsistency between statistical independence and semantic disentanglement. Experimental results on the dSprites and 3DShapes datasets demonstrate that the proposed DiD outperforms existing mainstream methods across various disentanglement metrics.