When Does Domain Adaptation Help on Physical Vibration Sensors? A Held-Out-Bearing Study of Neural-Operator and Convolutional Models

This paper revisits rolling-element bearing fault diagnosis from vibration under domain shift caused by different operating conditions. Unlike evaluations that place the same physical bearing in both training and test sets, the study uses a held-out-bearing protocol with each bearing assigned entirely to one split. It compares neural-operator and convolutional models in this physical-sensing setting.