发表机构
Liverpool John Moores University; University of Texas at Arlington; IIT Patna; Stony Brook University; IIT Indore; UMass Amherst（利物浦约翰摩尔大学; 德克萨斯大学阿灵顿分校; 印度理工学院巴特那分校; 石溪大学; 印度理工学院印多尔分校; 马萨诸塞大学阿默斯特分校）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出利用大型语言模型（LLMs）微调并结合检索增强生成（RAG）技术，为制造业生成合成时间序列数据，以解决标记数据稀缺问题，实验证明该方法在捕捉时间依赖性和提升异常检测等下游任务性能上优于ARIMA和LSTM等传统模型。
AI 中文摘要
本文提出了一种新颖的框架，利用大型语言模型（LLMs）为制造过程生成合成时间序列数据。鉴于真实制造环境中标记时间序列数据的稀缺性，这阻碍了稳健机器学习模型的发展，我们探索了LLMs学习复杂时间依赖关系并生成逼真合成数据的潜力。我们的方法包括在制造过程指令上微调预训练的LLMs，并采用检索增强生成（RAG）技术来增强数据的多样性和真实性。我们使用定量指标、PCA分析和下游任务性能（异常检测）将我们的方法与ARIMA和LSTM等传统时间序列建模技术进行评估。结果表明，我们基于LLM的框架优于这些基线，生成的高质量合成时间序列数据有效捕捉了真实制造数据的时间依赖性和统计特性，从而提升了下游任务性能。
英文摘要
This paper presents a novel framework leveraging Large Language Models (LLMs) to generate synthetic time series data for manufacturing processes. Motivated by the scarcity of labeled time-series data in real-world manufacturing settings, which hinders the development of robust machine learning models, we explore the potential of LLMs to learn complex temporal dependencies and generate realistic synthetic data. Our approach involves fine-tuning pre-trained LLMs on manufacturing process instructions and employing a Retrieval Augmented Generation (RAG) technique to enhance data diversity and realism. We evaluate our method against traditional time series modeling techniques like ARIMA and LSTMs, using quantitative metrics, PCA analysis, and downstream task performance (anomaly detection). Results demonstrate that our LLM-driven framework outperforms these baselines, generating high-quality synthetic time series data that effectively captures temporal dependencies and statistical properties of real manufacturing data, leading to improvements in downstream task performance.
Comments7 pages, 4 figures. Published in the 2024 International Conference on Image Processing, Computer Vision and Machine Learning (ICICML)
Journal ref2024 International Conference on Image Processing, Computer Vision and Machine Learning (ICICML), pp. 2053-2059, IEEE, 2024
DOI：10.1109/ICICML63543.2024.10958017
