Expert-Data Alignment Governs Generation Quality in Decentralized Diffusion Models

2026-02-02

AI Summary

This work challenges the prevailing assumption in decentralized diffusion models (DDMs) that numerical stability governs generation quality, arguing instead that quality is primarily determined by the alignment between experts and the current denoising state of the data. Through systematic analysis of data cluster distances, expert prediction accuracy, and inter-expert disagreement, the study compares sparse Top-2 routing against full ensemble routing mechanisms. Experiments on two DDM systems demonstrate that sparse Top-2 routing achieves a Fréchet Inception Distance (FID) of 22.6, substantially outperforming the full ensemble’s FID of 47.9. These results establish, for the first time, that expert–data alignment is the key principle underlying high-quality generation and introduce a new paradigm favoring sparse routing over full integration in DDM architectures.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Web data quality in the era of algorithmically-generated content

Abstract

Decentralized Diffusion Models (DDMs) route denoising through experts trained independently on disjoint data clusters, which can strongly disagree in their predictions. What governs the quality of generations in such systems? We present the first ever systematic investigation of this question. A priori, the expectation is that minimizing denoising trajectory sensitivity -- minimizing how perturbations amplify during sampling -- should govern generation quality. We demonstrate this hypothesis is incorrect: a stability-quality dissociation. Full ensemble routing, which combines all expert predictions at each step, achieves the most stable sampling dynamics and best numerical convergence while producing the worst generation quality (FID 47.9 vs. 22.6 for sparse Top-2 routing). Instead, we identify expert-data alignment as the governing principle: generation quality depends on routing inputs to experts whose training distribution covers the current denoising state. Across two distinct DDM systems, we validate expert-data alignment using (i) data-cluster distance analysis, confirming sparse routing selects experts with data clusters closest to the current denoising state, and (ii) per-expert analysis, showing selected experts produce more accurate predictions than non-selected ones, and (iii) expert disagreement analysis, showing quality degrades when experts disagree. For DDM deployment, our findings establish that routing should prioritize expert-data alignment over numerical stability metrics.

Problem

• Decentralized Diffusion Models
• generation quality
• expert-data alignment
• denoising trajectory
• expert disagreement

Innovation

• expert-data alignment
• decentralized diffusion models
• routing strategy
• generation quality
• stability-quality dissociation