Blind Denoising Diffusion Models Offer Theoretical Framework Without Explicit Noise Conditioning
Researchers have developed a complete theoretical framework for 'blind denoising diffusion models' (BDDMs), a variant of standard diffusion models that removes the requirement to feed noise amplitude information into neural networks during training and sampling. Standard denoising diffusion models — widely used for generating images and other data — rely on ad hoc noise schedules and noise embeddings that lack rigorous theoretical justification. The work, accepted to the ICML 2025 FoGen workshop, could simplify diffusion model design while providing a principled foundation for a key component of modern generative AI.
Denoising diffusion models (DDMs) are among the most powerful methods for learning and generating data distributions, underpinning many state-of-the-art image and audio generation systems. However, a critical part of their pipeline — noise conditioning, which requires passing noise amplitude information into the neural network — has lacked theoretical grounding and forces practitioners to rely on unprincipled design choices such as ad hoc noise embeddings and sampling schedules. Researchers Aram-Alexandre Pooladian and collaborators propose 'blind denoising diffusion models' (BDDMs), where the noise amplitude is withheld from the network entirely during both training and sampling. They justify the correctness of this approach under an assumption that the underlying data distribution has low intrinsic dimensionality relative to the ambient space — a condition they connect to the classical statistical problem of estimating noise levels from a single noisy observation. The paper introduces an adaptive sampling scheme that is rigorously supported by this analysis and demonstrates empirically that BDDMs can match or benefit over standard DDMs. The 39-page paper, including 13 figures, was accepted to the ICML 2025 Foundations of Generative Models (FoGen) workshop.
What's missing
The low intrinsic dimensionality assumption is a sufficient condition for correctness but its tightness or necessity is not established; empirical comparisons are limited to benchmarks described in the paper and may not generalize to all domains or model scales; and the practical overhead or limitations of the adaptive scheme in large-scale production settings are not addressed.
What different sources said
- arXiv cs.LGCenter
Blind denoising diffusion models and the blessings of dimensionality
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.