Gaussian Processes now condition on natural language via diffusion models

Researchers have developed a novel method to condition Gaussian Processes (GPs) on virtually any type of information, including natural language. This approach establishes an equivalence between GPs and linear diffusion models, allowing predictive sampling to be treated as an ODE. The technique handles non-conjugate conditioning, such as natural language via large language models, for the first time, opening new possibilities for probabilistic modeling. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Enables richer probabilistic modeling by integrating diverse data types, including natural language, into Gaussian Processes.

RANK_REASON Academic paper introducing a new methodology for Gaussian Processes. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• Gaussian Processes
• Lachlan Astfalck
• large language models

• paper
• model release