Research map / Probabilistic and scientific modeling / Variational and generative sampling
Cauchy–Schwarz Regularized Autoencoder
Linh Tran, Maja Pantic, Marc Peter Deisenroth
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 KL regularizer with Cauchy-Schwarz regularizer | VAE (z=40) · Dynamic MNIST | test FID: 10.99 -> 7.88 | Improved |
| replace KL regularizer with Cauchy-Schwarz regularizer | VAE (z=40) · Static MNIST | test FID: 10.04 -> 7.60 | Improved |
| replace KL regularizer with Cauchy-Schwarz regularizer | VAE (z=40) · CIFAR10 | test FID: 11.36 -> 10.94 | Improved |
| add mixture-of-Gaussians prior (100 components) to CSRAE | CSRAE · Caltech101 | test FID: 17.15 -> 16.33 | Improved |
| add mixture prior (100 components) to CSRAE | CSRAE · Static MNIST | test FID: 7.60 -> 7.68 | No clear effect |
| add mixture prior (100 components) to CSRAE | CSRAE · CIFAR10 | test FID: 10.94 -> 10.65 | Improved |
| use MixtureCSRAE instead of VampPriorVAE for kNN on latent codes | MixtureCSRAE (z=20) · Dynamic MNIST (k=3) | kNN error (%): 3.98 -> 2.98 | Improved |
| use MixtureCSRAE instead of VampPriorVAE for kNN on latent codes | MixtureCSRAE (z=20) · Caltech101 (k=10) | kNN error (%): 38.52 -> 34.94 | Improved |
| use MixtureCSRAE instead of VampPriorVAE with 250 labels | semi-supervised VAE · Dynamic MNIST | classification error (%): 4.73 -> 4.12 | Improved |
| use MixtureCSRAE instead of VampPriorVAE with 4000 labels | semi-supervised VAE · Dynamic MNIST | classification error (%): 3.15 -> 2.75 | Improved |
| use MixtureCSRAE instead of VampPriorVAE on FERA2015 10% labels | semi-supervised VAE · FERA2015 | average F1: 23.07 -> 30.56 | Improved |
| increase labels from 10% to 25% for MixtureCSRAE on FERA2015 | MixtureCSRAE · FERA2015 | average F1: 30.56 -> 27.11 | Hurt |
Closest papers by meaning
- Variational Autoencoder with Implicit Optimal Priors (AAAI 2019 · similarity 0.93)
- Consistency Regularization for Variational Auto-Encoders (NeurIPS 2021 · similarity 0.92)
- Shape your Space: A Gaussian Mixture Regularization Approach to Deterministic Autoencoders (NeurIPS 2021 · similarity 0.91)
- Gaussian Process Prior Variational Autoencoders (NeurIPS 2018 · similarity 0.91)
- Marginalization is not Marginal: No Bad VAE Local Minima when Learning Optimal Sparse Representations (ICML 2023 · similarity 0.91)
- Improving Variational Autoencoders with Density Gap-based Regularization (NeurIPS 2022 · similarity 0.91)
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.
