Large Language Models for Limited Noisy Data: A Gravitational Wave Identification Study

2025-12-03

AI Summary

This study addresses the challenge of detecting astrophysical signals—such as gravitational waves—in real observational data characterized by strong non-Gaussian, non-stationary noise and severe scarcity of labeled examples. We investigate the modeling advantages of large language models (LLMs) over conventional neural networks in this low-resource regime. We propose a novel LLM fine-tuning paradigm for few-shot astronomical time-series classification that bypasses reliance on large-scale synthetic waveforms and instead learns discriminative temporal structures directly from authentic observations. Fine-tuned on only 90 real LIGO gravitational-wave events, our model achieves 97.4% detection accuracy—substantially outperforming CNN and LSTM baselines trained on the same limited dataset. Performance scales predictably with both model size and data quantity. This work constitutes the first empirical validation of LLMs’ strong generalization capability in high-noise, low-label astrophysical tasks, establishing a new paradigm for scientific discovery under extreme annotation scarcity.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Computer Vision: Large Vision Models

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

This work investigates whether large language models (LLMs) offer advantages over traditional neural networks for astronomical data processing, in regimes with non-Gaussian, non-stationary noise and limited labeled samples. Gravitational wave observations provide an suitable test case, using only 90 LIGO events, finetuned LLMs achieve 97.4% accuracy for identifying signals. Further experiments show that, in contrast to traditional networks that rely on large simulated datasets, additional simulated samples do not improve LLM performance, while scaling studies reveal predictable gains with increasing model size and dataset size. These results indicate that LLMs can extract discriminative structure directly from observational data and provide an efficient assessment for gravitational wave identification. The same strategy may extend to other astronomical domains with similar noise properties, such as radio or pulsar observations.

Problem

• Evaluating LLMs vs traditional networks for astronomical data with noisy, limited samples.
• Testing LLMs on gravitational wave identification using only 90 real LIGO events.
• Extending LLM efficiency to other astronomical domains with similar noise challenges.

Innovation

• LLMs achieve high accuracy with limited real gravitational wave data
• LLMs do not require large simulated datasets for performance
• LLMs extract discriminative structure directly from observational data