SONA: Learning Conditional, Unconditional, and Mismatching-Aware Discriminator
This work addresses a fundamental problem in conditional generative adversarial networks for researchers and practitioners, offering an incremental improvement over existing methods.
The paper tackles the challenge of conditional generation in deep generative models by proposing SONA, a novel discriminator design that integrates unconditional discrimination, matching-aware supervision, and adaptive weighting, resulting in superior sample quality and conditional alignment compared to state-of-the-art methods in class-conditional and text-to-image generation tasks.
Deep generative models have made significant advances in generating complex content, yet conditional generation remains a fundamental challenge. Existing conditional generative adversarial networks often struggle to balance the dual objectives of assessing authenticity and conditional alignment of input samples within their conditional discriminators. To address this, we propose a novel discriminator design that integrates three key capabilities: unconditional discrimination, matching-aware supervision to enhance alignment sensitivity, and adaptive weighting to dynamically balance all objectives. Specifically, we introduce Sum of Naturalness and Alignment (SONA), which employs separate projections for naturalness (authenticity) and alignment in the final layer with an inductive bias, supported by dedicated objective functions and an adaptive weighting mechanism. Extensive experiments on class-conditional generation tasks show that \ours achieves superior sample quality and conditional alignment compared to state-of-the-art methods. Furthermore, we demonstrate its effectiveness in text-to-image generation, confirming the versatility and robustness of our approach.