Jul 2026
Semi-Supervised Conditional Generative Learning through Stochastic Interpolation and Sufficient Representations
The theoretical results demonstrate that this method effectively mitigates the curse of dimensionality inherent in direct ambient-space generative modeling and derive non-asymptotic convergence rates proving that RepG significantly improves sample complexity.
Changyu Liu, Yuling Jiao, Jian Huang
· arXiv.org · 0 citations