Self Iterative Label Refinement via Robust Unlabeled Learning
This work addresses the challenge of reducing human supervision in LLM self-refinement for classification tasks, though it is incremental as it builds on existing unlabeled learning frameworks.
The paper tackles the problem of LLMs' reliance on costly high-quality feedback and their biases in self-refinement by introducing an iterative refinement pipeline using unlabeled learning to improve pseudo-labels for classification tasks, resulting in consistent performance gains over initial LLM outputs and state-of-the-art self-refinement methods across diverse datasets.
Recent advances in large language models (LLMs) have yielded impressive performance on various tasks, yet they often depend on high-quality feedback that can be costly. Self-refinement methods attempt to leverage LLMs' internal evaluation mechanisms with minimal human supervision; however, these approaches frequently suffer from inherent biases and overconfidence, especially in domains where the models lack sufficient internal knowledge, resulting in performance degradation. As an initial step toward enhancing self-refinement for broader applications, we introduce an iterative refinement pipeline that employs the Unlabeled-Unlabeled learning framework to improve LLM-generated pseudo-labels for classification tasks. By exploiting two unlabeled datasets with differing positive class ratios, our approach iteratively denoises and refines the initial pseudo-labels, thereby mitigating the adverse effects of internal biases with minimal human supervision. Evaluations on diverse datasets, including low-resource language corpora, patent classifications, and protein structure categorizations, demonstrate that our method consistently outperforms both initial LLM's classification performance and the self-refinement approaches by cutting-edge models (e.g., GPT-4o and DeepSeek-R1).