LLM-Powered Anomaly Detection in Longitudinal Texts via Functional PCA

This research explores a novel application of Large Language Models (LLMs) in conjunction with Functional Principal Component Analysis (FPCA) for anomaly detection in sparse, longitudinal text data. The combination of LLMs for feature extraction and FPCA for identifying deviations presents a promising approach.

• •The research proposes a method for anomaly detection in text using LLMs and FPCA.
• •The approach specifically addresses sparse and longitudinal text data.
• •The use of FPCA offers a potentially robust method for identifying unusual patterns.