Publication Details

On Sensitivity of Learning with Limited Labelled Data to the Effects of Randomness: Impact of Interactions and Systematic Choices

PECHER, B.; SRBA, I.; BIELIKOVÁ, M. On Sensitivity of Learning with Limited Labelled Data to the Effects of Randomness: Impact of Interactions and Systematic Choices. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. Miami: Association for Computational Linguistics, 2024. p. 522-556. ISBN: 979-8-8917-6164-3.
Czech title
O citlivosti učenia s obmedzenými označenými dátami na efekty náhodnosti: Vplyv interakcií a systematických rozhodnutí
Type
conference paper
Language
English
Authors
Keywords

NLP in resource-constrained settings, in-context learning, fine-tuning, meta-learning, sensitivity, effects of randomness, stability

Abstract

While learning with limited labelled data can effectively deal with a lack of labels, it is also sensitive to the effects of uncontrolled randomness introduced by so-called randomness factors (i.e., non-deterministic decisions such as choice or order of samples). We propose and formalise a method to systematically investigate the effects of individual randomness factors while taking the interactions (dependence) between them into consideration. To this end, our method mitigates the effects of other factors while observing how the performance varies across multiple runs. Applying our method to multiple randomness factors across in-context learning and fine-tuning approaches on 7 representative text classification tasks and meta-learning on 3 tasks, we show that: 1) disregarding interactions between randomness factors in existing works led to inconsistent findings due to incorrect attribution of the effects of randomness factors, such as disproving the consistent sensitivity of in-context learning to sample order even with random sample selection; and 2) besides mutual interactions, the effects of randomness factors, especially sample order, are also dependent on more systematic choices unexplored in existing works, such as number of classes, samples per class or choice of prompt format.

Published
2024
Pages
522–556
Proceedings
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
ISBN
979-8-8917-6164-3
Publisher
Association for Computational Linguistics
Place
Miami
DOI
BibTeX
@inproceedings{BUT193223,
  author="PECHER, B. and SRBA, I. and BIELIKOVÁ, M.",
  title="On Sensitivity of Learning with Limited Labelled Data to the Effects of Randomness: Impact of Interactions and Systematic Choices",
  booktitle="Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing",
  year="2024",
  pages="522--556",
  publisher="Association for Computational Linguistics",
  address="Miami",
  doi="10.18653/v1/2024.emnlp-main.32",
  isbn="979-8-8917-6164-3"
}
Back to top