MSDDA derives a closed-form optimal reverse denoising distribution for multi-objective diffusion alignment that is exactly equivalent to step-level RL fine-tuning with no approximation error.
Three-way trade-off in multi-objective learning: Optimization, generalization and conflict-avoidance.Advances in Neural Information Processing Systems, 36:70045–70093, 2023
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Step-level Denoising-time Diffusion Alignment with Multiple Objectives
MSDDA derives a closed-form optimal reverse denoising distribution for multi-objective diffusion alignment that is exactly equivalent to step-level RL fine-tuning with no approximation error.