Research map / Probabilistic and scientific modeling / Variational and generative sampling
Vector Quantization-Based Regularization for Autoencoders
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 |
|---|---|---|---|
| Use a soft quantized bottleneck with noise and a Bayesian estimator instead of the strongest listed baseline | Autoencoder latent representations · MNIST | clustering accuracy: 73.81 -> 77.64 | Helped |
| Use a soft quantized bottleneck with noise and a Bayesian estimator instead of the strongest listed baseline | Autoencoder latent representations · MNIST | clustering (NMI): 0.5760 -> 0.7188 | Helped |
| Use a soft quantized bottleneck with noise and a Bayesian estimator instead of raw input features | Autoencoder latent representations · MNIST | classification accuracy: 92.44 -> 93.54 | Helped |
| Use a soft quantized bottleneck with noise and a Bayesian estimator instead of the strongest listed baseline | Autoencoder latent representations · SVHN | clustering accuracy: 14.54 -> 17.68 | Helped |
| Use a soft quantized bottleneck with noise and a Bayesian estimator instead of the strongest listed baseline | Autoencoder latent representations · SVHN | classification accuracy: 49.62 -> 50.48 | Helped |
| Use a soft quantized bottleneck with noise and a Bayesian estimator instead of the strongest listed baseline | Autoencoder latent representations · CIFAR-10 | clustering accuracy: 24.12 -> 23.83 | Hurt |
| Use a soft quantized bottleneck with noise and a Bayesian estimator instead of the strongest listed baseline | Autoencoder latent representations · CIFAR-10 | classification accuracy: 40.92 -> 44.54 | Helped |
| Use a larger codebook than K=32 (ablation) | Soft VQ-VAE on CIFAR-10 and SVHN · CIFAR-10 and SVHN | downstream task accuracy | Hurt |
Closest papers by meaning
- Hierarchical Quantized Autoencoders (NeurIPS 2020 · similarity 0.89)
- Variational Bayesian Quantization (ICML 2020 · similarity 0.89)
- Gradient $\ell_1$ Regularization for Quantization Robustness (ICLR 2020 · similarity 0.89)
- Information-theoretic Generalization Analysis for VQ-VAEs: A Role of Latent Variables (NeurIPS 2025 · similarity 0.88)
- Restructuring Vector Quantization with the Rotation Trick (ICLR 2025 · similarity 0.88)
- Learning to Quantize for Training Vector-Quantized Networks (ICML 2025 · similarity 0.87)
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.