GOTabPFN: New Method for Applying Foundation Models to High-Dimensional Tabular Data
Researchers have introduced GOTabPFN, a method that enhances small tabular foundation models for high-dimensional, low-sample-size (HDLSS) prediction tasks without retraining large backbones. The approach combines Graph-guided Ordering with Local Refinement (GO-LR) and a Neuro-Inspired Subunit Compression (NSC) unit to create compact feature representations within tight computational budgets. The work, accepted to ICML 2026, addresses a practical bottleneck in applying transformer-style tabular models to domains like genomics or clinical data where samples are scarce but features are numerous.
GOTabPFN is a new framework designed to make TabPFN-style tabular foundation models practical in settings where datasets have many features but few samples — a common challenge in biomedical, scientific, and industrial applications. The core innovation is Graph-guided Ordering with Local Refinement (GO-LR), a feature-ordering strategy the authors show is mathematically equivalent to weighted Minimum Linear Arrangement, with the practical solver interpreted as a traveling-salesman-path-style surrogate. Locally ordered features are then pooled into compact 'meta-features' by a Neuro-Inspired Subunit Compression (NSC) unit, reducing the token count required for prediction. Crucially, the method works without retraining the underlying large backbone model, making it computationally accessible. Across multiple tabular benchmarks, GOTabPFN demonstrated improved stability and accuracy under tight token budgets compared to baseline approaches. The paper was accepted to the 43rd International Conference on Machine Learning (ICML 2026) and is accompanied by open-source code on GitHub, a PyPI package, and a Hugging Face demo.
What's missing
The abstract does not specify which benchmark datasets were used for evaluation, the magnitude of accuracy improvements over baselines, or how GOTabPFN compares to other HDLSS-specific methods beyond TabPFN variants. Scalability limits (e.g., maximum feature dimensionality tested) and potential failure modes are not discussed in the available summary.
What different sources said
- arXiv cs.AICenter
GOTabPFN: From Feature Ordering to Compact Tokenization for Tabular Foundation Models on High-Dimensional Data
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.