Research map / Probabilistic and scientific modeling / Variational and generative sampling
Adversarially Regularized 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 |
|---|---|---|---|
| Replace a fixed prior with a learned adversarial prior | ARAE text generation · Text sample generation | Reverse PPL: 980 -> 82.2 | Helped |
| Use ARAE instead of an AE | Text sample generation · Text sample generation | Reverse PPL: 97.3 -> 82.2 | Helped |
| Use ARAE instead of an AE | Text sample generation · Text sample generation | Forward PPL: 87.8 -> 44.3 | Helped |
| Increase adversarial-loss weight from λ¹ₐ to λ¹ᵦ (ablation) | ARAE sentiment transfer · Yelp sentiment transfer | Transfer: 73.4% -> 81.8% | Helped |
| Increase adversarial-loss weight from λ¹ₐ to λ¹ᵦ (ablation) | ARAE sentiment transfer · Yelp sentiment transfer | BLEU: 31.15 -> 20.18 | Hurt |
| Add adversarial regularization to the AE | Semi-supervised SNLI model · SNLI test, Medium | Accuracy: 68.5% -> 70.9% | Helped |
| Add adversarial regularization to the AE | Semi-supervised SNLI model · SNLI test, Small | Accuracy: 64.6% -> 66.8% | Helped |
| Add adversarial regularization to the AE | Semi-supervised SNLI model · SNLI test, Tiny | Accuracy: 59.9% -> 62.5% | Helped |
| Train on additional unlabeled data with an AE objective | SNLI supervised encoder · SNLI test | Accuracy: 65.9% -> 68.5%; 62.5% -> 64.6%; 57.9% -> 59.9% | Helped |
| Increase adversarial regularization (ablation) | ARAE sentiment transfer · Yelp sentiment transfer | Similarity: 3.8 -> 3.7 | Hurt |
Closest papers by meaning
- Educating Text Autoencoders: Latent Representation Guidance via Denoising (ICML 2020 · similarity 0.90)
- Wasserstein Auto-Encoders (ICLR 2018 · similarity 0.88)
- On Variational Learning of Controllable Representations for Text without Supervision (ICML 2020 · similarity 0.88)
- Statistical Regeneration Guarantees of the Wasserstein Autoencoder with Latent Space Consistency (NeurIPS 2021 · similarity 0.88)
- StrWAEs to Invariant Representations (ICML 2024 · similarity 0.86)
- A Statistical Analysis of Wasserstein Autoencoders for Intrinsically Low-dimensional Data (ICLR 2024 · similarity 0.86)
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.