CRUMB: New Method Speeds Up Prior-Fitted Network Inference for Large Datasets
Researchers have proposed CRUMB, a three-stage inference wrapper designed to make Prior-Fitted Networks (PFNs) scalable to large tabular datasets without retraining. PFNs are a class of tabular foundation models that struggle with quadratically scaling attention costs when training sets are very large. CRUMB addresses this bottleneck by intelligently selecting small, distributionally matched data subsets, potentially broadening the practical applicability of PFN-based models.
Prior-Fitted Networks (PFNs) perform in-context learning by taking an entire labelled training set as input and producing predictions in a single forward pass, but their self-attention mechanism scales quadratically with dataset size, making inference on large datasets computationally prohibitive. To address this, researchers introduce CRUMB (Clustered Retrieval Using Minimised-MMD Batching), which operates in three stages: clustering test queries, selecting a distributionally matched training subset for each cluster by greedily minimising the Maximum Mean Discrepancy (MMD), and then running standard PFN inference on each reduced-context batch. Crucially, CRUMB is architecture-agnostic and requires no retraining of the underlying model, making it a drop-in inference wrapper. Evaluated on the 51-dataset TabArena benchmark across three PFN architectures—TabPFNv2, TabICLv1, and TabICLv2—CRUMB outperforms comparable state-of-the-art context selection strategies. The MMD-minimisation step also provides a natural robustness to covariate drift by aligning the training context distribution with the test batch distribution.
What's missing
The paper does not report wall-clock inference time comparisons or memory usage benchmarks against baselines, which would clarify the practical computational savings. It is also unclear how CRUMB performs on datasets with severe class imbalance or very high dimensionality, and whether the greedy MMD minimisation introduces any worst-case approximation guarantees.
What different sources said
- arXiv cs.AICenter
CRUMB: Efficient Prior Fitted Network Inference via Distributionally Matched Context Batching
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.