Geometric Filtering of LLM-Generated Samples for Few-Shot Text Classification

A new geometric filtering framework for selecting LLM-generated training samples improves few-shot text classification by +2.61 percentage points over SMOTE across 13 datasets, 5 classifiers, 10 augmentation methods, and over 6,700 configurations, with a 88.9% win rate and Cohen's d=0.95 (p<0.0001). The method also generalizes to named entity recognition (+9.26pp, 100% win rate) without modification and is robust across 5 LLMs from 4 providers, according to the arXiv preprint 2608.13866v1.

arXiv:2608.13866v1 Announce Type: new Abstract: Large language models LLMs can generate synthetic training data for text classification, but the quality of generated samples is heterogeneous: some fall in correct class regions of the embedding space while others land in peripheral or cross-class zones. We propose a geometric filtering framework that evaluates each LLM-generated sample by its Euclidean distance to real class examples in a sentence embedding space, selecting only geometrically consistent candidates. A soft weighting mechanism transforms filter scores into sample weights for classifier training. Evaluated across 13 datasets, 5 classifiers, 10 augmentation methods, and over 6,700 configurations, our method achieves +2.61 percentage points pp over SMOTE $p<0.0001$, Cohen's $d=0.95$, 88.9% win rate . The approach generalizes to named entity recognition +9.26pp, 100% win rate without filter modification, and is robust across 5 LLMs from 4 providers. A key finding is that the simplest distance-based filter consistently outperforms complex multi-criteria alternatives.