Research map / Learning theory and algorithms / Sample complexity
Wasserstein Dependency Measure for Representation Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Replace CPC with WPC using a fully connected network | Representation learning on SplitCelebA · SplitCelebA | linear-probe accuracy: 0.85 -> 0.87 | Helped |
| Replace CPC with WPC using a convolutional network | Representation learning on SplitCelebA · SplitCelebA | linear-probe accuracy: 0.82 -> 0.87 | Helped |
| Replace CPC with WPC | Representation learning on StackedMultiOmniglot with fully connected networks · StackedMultiOmniglot | linear-probe performance | Helped |
| Replace CPC with WPC | Representation learning on MultiviewShapes3D with fully connected networks · MultiviewShapes3D | linear-probe performance | Helped |
| Increase the number of characters, raising mutual information beyond what the dataset size supports | CPC representation learning · StackedMultiOmniglot | linear-probe performance | Hurt |
| Increase the number of characters and mutual information | WPC representation learning · StackedMultiOmniglot | linear-probe performance | Hurt |
| Use WPC instead of CPC | Representation learning on SpatialMultiOmniglot with convolutional networks · SpatialMultiOmniglot | linear-probe performance | Helped |
| Use WPC instead of CPC | Representation learning on StackedMultiOmniglot with convolutional networks · StackedMultiOmniglot | linear-probe performance | Helped |
| Use WPC instead of CPC | Representation learning on MultiviewShapes3D with convolutional networks · MultiviewShapes3D | linear-probe performance | Helped |
| Increase minibatch size beyond 32 | WPC representation learning · SpatialMultiOmniglot | linear-probe performance | No clear effect |
Closest papers by meaning
- On Mutual Information Maximization for Representation Learning (ICLR 2020 · similarity 0.87)
- Learning Wasserstein Embeddings (ICLR 2018 · similarity 0.86)
- Connecting Jensen–Shannon and Kullback–Leibler Divergences: A New Bound for Representation Learning (NeurIPS 2025 · similarity 0.86)
- Generalization Guarantees for Representation Learning via Data-Dependent Gaussian Mixture Priors (ICLR 2025 · similarity 0.85)
- Fourier Sliced-Wasserstein Embedding for Multisets and Measures (ICLR 2025 · similarity 0.84)
- Towards Principled Representation Learning from Videos for Reinforcement Learning (ICLR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.