SONA: Learning Conditional, Unconditional, and Mismatching-Aware Discriminator

2025-10-06

AI Summary

Existing conditional GAN discriminators struggle to jointly optimize sample fidelity and conditional alignment. To address this, we propose a novel conditional discriminator featuring a decoupled architecture: an unconditional branch evaluates sample authenticity, while a matching-aware branch models conditional consistency; separation via projection-based structures and inductive biases further enhances feature disentanglement. We additionally introduce matching-aware supervision, adaptive loss weighting, and a dual-branch output mechanism to dynamically balance the two objectives during training. Our method achieves significant improvements in FID and CLIP Score on both class-conditional and text-to-image generation benchmarks, consistently outperforming state-of-the-art approaches in both sample quality and conditional alignment. Comprehensive experiments validate its effectiveness, robustness, and generalizability across diverse conditional generation tasks.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

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.

Problem

• Balancing authenticity and alignment in conditional generation
• Improving conditional discriminator design with three capabilities
• Enhancing sample quality and alignment in generative tasks

Innovation

• Separate projections for naturalness and alignment
• Adaptive weighting mechanism for dynamic balance
• Matching-aware supervision enhances alignment sensitivity