Research map / Image generation / GANs
Invertibility of Convolutional Generative Networks from Partial Measurements
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 ReLU with Sigmoid or Tanh activation (ablation) | 2-layer transposed-conv generator inversion · simulated 2-layer network, 2-D latent | cost landscape (unique global minimum) | Helped |
| replace ReLU with Leaky ReLU (ablation) | 2-layer transposed-conv generator inversion · simulated 2-layer network | cost landscape | No clear effect |
| draw kernel weights uniformly from [0,1] instead of zero-mean Gaussian (ablation) | 2-layer generator · simulated 2-layer network | invertibility (unique global minimum) | Hurt |
| observe spatially dispersed samples versus contiguous halves of the image (ablation) | latent code recovery from partial measurements · sampling patterns | minimum sampling rate for perfect recovery | Helped |
| apply gradient descent inversion to multi-layer GANs trained on real data | MNIST and CelebA DCGAN generators · MNIST, CelebA | latent code recovery | Helped |
Closest papers by meaning
- Inverting Deep Generative models, One layer at a time (NeurIPS 2019 · similarity 0.91)
- i-RevNet: Deep Invertible Networks (ICLR 2018 · similarity 0.91)
- PLUGIn: A simple algorithm for inverting generative models with recovery guarantees (NeurIPS 2021 · similarity 0.90)
- Invertible generative models for inverse problems: mitigating representation error and dataset bias (ICML 2020 · similarity 0.90)
- Invertible Residual Networks (ICML 2019 · similarity 0.89)
- Network Inversion of Convolutional Neural Nets (Student Abstract) (AAAI 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.