RCAP: New Algorithm for Efficient Model Training Through Intelligent Data Selection
Researchers have proposed RCAP, a dynamic dataset pruning algorithm that selects representative training subsets on a class-aware, probabilistic basis to reduce computational cost while preserving model accuracy. Unlike existing pruning methods, RCAP adaptively adjusts per-class sample fractions each epoch using class-wise loss signals and prioritizes high-loss samples. The approach is notable for maintaining strong worst-group accuracy even at aggressive pruning rates, which is critical for fairness in models trained on imbalanced real-world data.
RCAP (Robust, Class-Aware, Probabilistic) is a dynamic dataset pruning algorithm introduced for classification tasks, accepted at the Forty-first Conference on Uncertainty in Artificial Intelligence (UAI 2025). The method addresses a known weakness in existing pruning techniques: degraded worst-group accuracy, especially at high pruning rates and on imbalanced datasets. RCAP uses a closed-form solution to estimate per-class inclusion fractions, updating them every epoch based on aggregated class-wise loss, and then applies an adaptive sampling strategy that favors high-loss samples within each class. The authors evaluated RCAP across six datasets spanning balanced to highly imbalanced distributions, five model architectures, and three training paradigms—training from scratch, transfer learning, and fine-tuning. With only 10% of training data, RCAP reportedly exceeds full-data training performance by more than 1% on class-imbalanced datasets while delivering an average 8.69× training speedup. The method consistently outperformed state-of-the-art baselines across all tested pruning rates.
What's missing
The study does not report statistical significance measures or confidence intervals for the performance gains, making it difficult to assess whether the improvements are robust across random seeds and initialization conditions. The evaluation is limited to classification tasks, and generalizability to other problem types (e.g., regression, generation) is not addressed.
What different sources said
- arXiv cs.LGCenter
RCAP: Robust, Class-Aware, Probabilistic Dynamic Dataset Pruning
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.