Xianguang Li

A Task-Based Few-Shot Learning Framework for Wind Turbine Gearbox Fault Diagnosis via Large Language Models

Inspired by few-shot learning, a support–query task construction strategy is introduced, reformulating classification as conditional task reasoning and a global coverage and local balance task sampling strategy is designed to enhance task diversity and mitigate sample imbalance.