Beyond Distribution Matching: Semantics-Consistent Tabular Diffusion with Weak Semantic Priors
Extensive experiments on six real-world tabular benchmarks show that the proposed semantics-consistent tabular diffusion framework consistently improves distributional fidelity, semantic consistency, and downstream task utility over representative VAE, GAN, LLM, and diffusion-based baselines.
Yi-Li Wang, Ru-Xue Shi, Mengnan Du et al.
· 0 citations