Saved in:
Bibliographic Details
Main Authors: Rahman, Md Mezbaur, Caragea, Cornelia
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.16516
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914051941990400
author Rahman, Md Mezbaur
Caragea, Cornelia
author_facet Rahman, Md Mezbaur
Caragea, Cornelia
contents In this paper, we introduce a novel weighted co-training approach that is guided by Large Language Models (LLMs). Namely, in our co-training approach, we use LLM labels on unlabeled data as target labels and co-train two encoder-only based networks that train each other over multiple iterations: first, all samples are forwarded through each network and historical estimates of each network's confidence in the LLM label are recorded; second, a dynamic importance weight is derived for each sample according to each network's belief in the quality of the LLM label for that sample; finally, the two networks exchange importance weights with each other -- each network back-propagates all samples weighted with the importance weights coming from its peer network and updates its own parameters. By strategically utilizing LLM-generated guidance, our approach significantly outperforms conventional SSL methods, particularly in settings with abundant unlabeled data. Empirical results show that it achieves state-of-the-art performance on 4 out of 5 benchmark datasets and ranks first among 14 compared methods according to the Friedman test. Our results highlight a new direction in semi-supervised learning -- where LLMs serve as knowledge amplifiers, enabling backbone co-training models to achieve state-of-the-art performance efficiently.
format Preprint
id arxiv_https___arxiv_org_abs_2509_16516
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LLM-Guided Co-Training for Text Classification
Rahman, Md Mezbaur
Caragea, Cornelia
Machine Learning
In this paper, we introduce a novel weighted co-training approach that is guided by Large Language Models (LLMs). Namely, in our co-training approach, we use LLM labels on unlabeled data as target labels and co-train two encoder-only based networks that train each other over multiple iterations: first, all samples are forwarded through each network and historical estimates of each network's confidence in the LLM label are recorded; second, a dynamic importance weight is derived for each sample according to each network's belief in the quality of the LLM label for that sample; finally, the two networks exchange importance weights with each other -- each network back-propagates all samples weighted with the importance weights coming from its peer network and updates its own parameters. By strategically utilizing LLM-generated guidance, our approach significantly outperforms conventional SSL methods, particularly in settings with abundant unlabeled data. Empirical results show that it achieves state-of-the-art performance on 4 out of 5 benchmark datasets and ranks first among 14 compared methods according to the Friedman test. Our results highlight a new direction in semi-supervised learning -- where LLMs serve as knowledge amplifiers, enabling backbone co-training models to achieve state-of-the-art performance efficiently.
title LLM-Guided Co-Training for Text Classification
topic Machine Learning
url https://arxiv.org/abs/2509.16516