Research map / Image generation / GANs
GibbsNet: Iterative Adversarial Inference for Deep Graphical Models
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 ALI with GibbsNet while keeping the inference, generation, and discriminator architectures the same | ALI · Permutation invariant MNIST test | test accuracy: 91% -> 97.7% | Helped |
| Replace ALI with GibbsNet while keeping the inference, generation, and discriminator architectures the same | ALI · SVHN test | test accuracy: 66.7% -> 79.6% | Helped |
| Replace ALI with GibbsNet | ALI · CIFAR-10 samples | Inception Score: 5.41 -> 5.69 | Helped |
| Replace ALI with GibbsNet | ALI · CIFAR-10 inpainting | Inception Score: 5.59 -> 6.15 | Helped |
| Increase ALI inference-network depth from 2 to 8 fully-connected layers (ablation) | ALI · MNIST validation | classification accuracy: 89.4% to 91.0% | No clear effect |
| Use a 6-layer inference network with 2048 units per layer and a 256-dimensional latent prior (ablation) | ALI · MNIST validation | classification accuracy: 91.2% | No clear effect |
| Use a stochastic rather than deterministic decoder (ablation) | GibbsNet · MNIST digits 0, 1, and 9 latent-space visualization | latent-space class separation | Mixed |
| Give latent variables to the discriminator (ablation) | GibbsNet · Inference | inference quality | Helped |
Closest papers by meaning
- Generalized Adversarially Learned Inference (AAAI 2021 · similarity 0.89)
- Adversarially-learned Inference via an Ensemble of Discrete Undirected Graphical Models (NeurIPS 2020 · similarity 0.89)
- Adversarially Learned Inference (ICLR 2017 · similarity 0.87)
- Neural Snowflakes: Universal Latent Graph Inference via Trainable Latent Geometries (ICLR 2024 · similarity 0.86)
- Unlocking Dataset Distillation with Diffusion Models (NeurIPS 2025 · similarity 0.86)
- On the Stability of Graph Convolutional Neural Networks: A Probabilistic Perspective (NeurIPS 2025 · similarity 0.85)
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.