Saved in:
Bibliographic Details
Main Authors: Ogonowski, Aleksander, Klimaszewski, Konrad, Rokita, Przemysław
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.17637
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915871895584768
author Ogonowski, Aleksander
Klimaszewski, Konrad
Rokita, Przemysław
author_facet Ogonowski, Aleksander
Klimaszewski, Konrad
Rokita, Przemysław
contents We present DSS-GAN, the first generative adversarial network to employ Mamba as a hierarchical generator backbone for noise-to-image synthesis. The central contribution is Directional Latent Routing (DLR), a novel conditioning mechanism that decomposes the latent vector into direction-specific subvectors, each jointly projected with a class embedding to produce a feature-wise affine modulation of the corresponding Mamba scan. Unlike conventional class conditioning that injects a global signal, DLR couples class identity and latent structure along distinct spatial axes of the feature map, applied consistently across all generative scales. DSS-GAN achieves improved FID, KID, and precision-recall scores compared to StyleGAN2-ADA across multiple tested datasets. Analysis of the latent space reveals that directional subvectors exhibit measurable specialization: perturbations along individual components produce structured, direction-correlated changes in the synthesized image.
format Preprint
id arxiv_https___arxiv_org_abs_2603_17637
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle DSS-GAN: Directional State Space GAN with Mamba backbone for Class-Conditional Image Synthesis
Ogonowski, Aleksander
Klimaszewski, Konrad
Rokita, Przemysław
Machine Learning
Computer Vision and Pattern Recognition
We present DSS-GAN, the first generative adversarial network to employ Mamba as a hierarchical generator backbone for noise-to-image synthesis. The central contribution is Directional Latent Routing (DLR), a novel conditioning mechanism that decomposes the latent vector into direction-specific subvectors, each jointly projected with a class embedding to produce a feature-wise affine modulation of the corresponding Mamba scan. Unlike conventional class conditioning that injects a global signal, DLR couples class identity and latent structure along distinct spatial axes of the feature map, applied consistently across all generative scales. DSS-GAN achieves improved FID, KID, and precision-recall scores compared to StyleGAN2-ADA across multiple tested datasets. Analysis of the latent space reveals that directional subvectors exhibit measurable specialization: perturbations along individual components produce structured, direction-correlated changes in the synthesized image.
title DSS-GAN: Directional State Space GAN with Mamba backbone for Class-Conditional Image Synthesis
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.17637