Style Quantization for Data-Efficient GAN Training
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916667990212608 |
|---|---|
| author | Wang, Jian Lan, Xin Zhou, Jizhe Tian, Yuxin Lv, Jiancheng |
| author_facet | Wang, Jian Lan, Xin Zhou, Jizhe Tian, Yuxin Lv, Jiancheng |
| contents | Under limited data setting, GANs often struggle to navigate and effectively exploit the input latent space. Consequently, images generated from adjacent variables in a sparse input latent space may exhibit significant discrepancies in realism, leading to suboptimal consistency regularization (CR) outcomes. To address this, we propose \textit{SQ-GAN}, a novel approach that enhances CR by introducing a style space quantization scheme. This method transforms the sparse, continuous input latent space into a compact, structured discrete proxy space, allowing each element to correspond to a specific real data point, thereby improving CR performance. Instead of direct quantization, we first map the input latent variables into a less entangled ``style'' space and apply quantization using a learnable codebook. This enables each quantized code to control distinct factors of variation. Additionally, we optimize the optimal transport distance to align the codebook codes with features extracted from the training data by a foundation model, embedding external knowledge into the codebook and establishing a semantically rich vocabulary that properly describes the training dataset. Extensive experiments demonstrate significant improvements in both discriminator robustness and generation quality with our method. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_24282 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Style Quantization for Data-Efficient GAN Training Wang, Jian Lan, Xin Zhou, Jizhe Tian, Yuxin Lv, Jiancheng Computer Vision and Pattern Recognition Under limited data setting, GANs often struggle to navigate and effectively exploit the input latent space. Consequently, images generated from adjacent variables in a sparse input latent space may exhibit significant discrepancies in realism, leading to suboptimal consistency regularization (CR) outcomes. To address this, we propose \textit{SQ-GAN}, a novel approach that enhances CR by introducing a style space quantization scheme. This method transforms the sparse, continuous input latent space into a compact, structured discrete proxy space, allowing each element to correspond to a specific real data point, thereby improving CR performance. Instead of direct quantization, we first map the input latent variables into a less entangled ``style'' space and apply quantization using a learnable codebook. This enables each quantized code to control distinct factors of variation. Additionally, we optimize the optimal transport distance to align the codebook codes with features extracted from the training data by a foundation model, embedding external knowledge into the codebook and establishing a semantically rich vocabulary that properly describes the training dataset. Extensive experiments demonstrate significant improvements in both discriminator robustness and generation quality with our method. |
| title | Style Quantization for Data-Efficient GAN Training |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2503.24282 |