SyncSDE: A Probabilistic Framework for Diffusion Synchronization
This work addresses the problem of improving collaborative generation for AI researchers and practitioners, but it appears incremental as it builds on existing synchronization methods with task-specific optimizations.
The paper tackles the problem of suboptimal performance in collaborative generation using multiple diffusion models due to naive heuristics like averaging, by developing a probabilistic framework to analyze and optimize diffusion synchronization. It achieves better results by identifying optimal correlation models per task, though no concrete numbers are provided.
There have been many attempts to leverage multiple diffusion models for collaborative generation, extending beyond the original domain. A prominent approach involves synchronizing multiple diffusion trajectories by mixing the estimated scores to artificially correlate the generation processes. However, existing methods rely on naive heuristics, such as averaging, without considering task specificity. These approaches do not clarify why such methods work and often produce suboptimal results when a heuristic suitable for one task is blindly applied to others. In this paper, we present a probabilistic framework for analyzing why diffusion synchronization works and reveal where heuristics should be focused; modeling correlations between multiple trajectories and adapting them to each specific task. We further identify optimal correlation models per task, achieving better results than previous approaches that apply a single heuristic across all tasks without justification.