Verified80% confidenceFactExact time
The theoretical foundations of diffusion models can be traced back to a 2015 paper on non-equilibrium thermodynamics published by Sohl-Dickstein et al.
9
Sources
80%
Confidence
Long-term
Relevance
7/7/2026
First Seen
Sources
AI造假产业链揭秘：从Lily Jay事件看深度伪造与身份欺骗
hackernewshackernews7/4/2026
Related Claims
VerifiedIn 2015, Sohl-Dickstein et al. first introduced the diffusion process into generative models.74% similarUnverifiedThe theoretical foundations of diffusion models can be traced back to 2015 when Sohl-Dickstein et al. first introduced diffusion processes into deep learning, with a breakthrough in image generation achieved after Ho et al. proposed DDPM in 2020.67% similarUnverified对抗性样本（Adversarial Examples）的理论基础由Szegedy等人于2013年首次系统阐述64% similarUnverifiedThe monograph The Principles of Diffusion Models was authored by Lai et al.60% similarUnverifiedThe research introduces the high-temperature Marchenko-Pastur (HTMP) distribution assumption to characterize heavy-tailed properties.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/203334API
curl https://kongchang.com/api/v1/knowledge/claims/203334MCP
get_claim(id=203334)