Mixing Synthetic Data Sources Outperforms Selecting Single Generators for Time Series Foundation Models
Researchers found that combining multiple synthetic data generators equally produces better results than selecting the single best generator for training time series foundation models. The study evaluated 11 generator families across two model architectures and discovered that generator rankings vary by architecture, making selection unreliable. This finding reframes synthetic data pretraining as a corpus composition problem rather than a generator selection problem.
A new study accepted at the ICML 2026 Workshop on Foundation Models for Structured Data demonstrates that equal-weight mixtures of synthetic data generators consistently match or exceed the performance of individually selected best generators for time series foundation model pretraining. The researchers evaluated 11 generator families on two architectures—Chronos-T5-Mini and Moirai-Small—and found that the best and worst generators can produce up to a 2× gap in forecasting error under identical training budgets. Critically, generator rankings are not stable across different model architectures, meaning a top-performing generator for one architecture may underperform for another. Rather than solving the difficult generator selection problem, the authors propose sidestepping it entirely by combining all generators equally, an approach that yields stronger pretraining corpora when mixed with real data. The findings suggest that composition choices should be validated per model family rather than assumed to transfer across architectures.
What's missing
The study does not discuss computational costs of training with mixed synthetic corpora versus single generators, nor does it address scalability to larger model families or real-world deployment considerations. The paper also does not explore whether the equal-weight mixing strategy remains optimal as the number of generators increases or how sensitive results are to the specific generator families included.
What different sources said
- arXiv cs.LGCenter
Mix, Don't Pick: Why Synthetic Corpus Composition Matters for Time Series Foundation Model Pretraining
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.