Research map / Image generation / GANs
Multi-Task Adversarial Network for Disentangled Feature Learning
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 |
|---|---|---|---|
| Add adversarial content-style training and image generation (ablation) | Font recognition representation · Unseen Font | recognition rate (%): 36.5 -> 47.9 | Helped |
| Add adversarial content-style training and image generation (ablation) | Font recognition representation · Unseen Glyph | recognition rate (%): 38.6 -> 52.8 | Helped |
| Add adversarial content-style training and image generation (ablation) | Font recognition representation · Unseen Font and Glyph | recognition rate (%): 22.9 -> 34.8 | Helped |
| Replace cooperative content-and-style multitask training with adversarial training (ablation) | Font recognition representation · Unseen Font | recognition rate (%): 24.3 -> 43.2 | Helped |
| Replace cooperative content-and-style multitask training with adversarial training (ablation) | Font recognition representation · Unseen Glyph | recognition rate (%): 42.3 -> 49.5 | Helped |
| Add image generation to adversarial content-style training (ablation) | Font recognition representation · Unseen Font | recognition rate (%): 43.2 -> 47.9 | Helped |
| Add image generation to adversarial content-style training (ablation) | Font recognition representation · Unseen Glyph | recognition rate (%): 49.5 -> 52.8 | Helped |
| Add image generation to adversarial content-style training (ablation) | Font recognition representation · Unseen Font and Glyph | recognition rate (%): 30.4 -> 34.8 | Helped |
| Use multi-class EMD instead of cross-entropy for style adversarial training (ablation) | MTAN font model · Font test set | font recognition accuracy | Helped |
| Use multi-class EMD instead of cross-entropy for style adversarial training (ablation) | MTAN font model · Font test set | glyph recognition accuracy | Helped |
| Add adversarial training (ablation) | MTAN face model · Multi-PIE Avg. | recognition rate (%): 79.3 -> 88.9 | Helped |
| Add illumination as an additional factor to disentangle (ablation) | MTAN face model · Multi-PIE Avg. | recognition rate (%): 88.9 -> 89.6 | Helped |
Closest papers by meaning
- Disentangled Representation Learning GAN for Pose-Invariant Face Recognition (CVPR 2017 · similarity 0.87)
- Feature-Level Frankenstein: Eliminating Variations for Discriminative Recognition (CVPR 2019 · similarity 0.86)
- Overcoming the Disentanglement vs Reconstruction Trade-off via Jacobian Supervision (ICLR 2019 · similarity 0.86)
- Exploring Disentangled Feature Representation Beyond Face Identification (CVPR 2018 · similarity 0.85)
- Multi-Attribute Transfer via Disentangled Representation (AAAI 2019 · similarity 0.85)
- Masked Face Recognition with Generative-to-Discriminative Representations (ICML 2024 · similarity 0.84)
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.