Weak Signal Learning Dataset and Baseline Method

This paper addresses the critical need for a dedicated dataset in weak signal learning (WSL), a challenging area due to noise and imbalance. The authors construct a specialized dataset and propose a novel model (PDVFN) to tackle the difficulties of low SNR and class imbalance. This work is significant because it provides a benchmark and a starting point for future research in WSL, particularly in fields like fault diagnosis and medical imaging where weak signals are prevalent.