Robustness and Sensitivity of BERT Models Predicting Alzheimer's Disease from Text
This addresses robustness and sensitivity issues for BERT models in Alzheimer's disease prediction, which is incremental as it focuses on analyzing existing methods rather than introducing new ones.
The paper analyzed how text alterations affect BERT models predicting Alzheimer's disease from text, finding that BERT is robust to natural linguistic variations but not sensitive to removing clinically important information.
Understanding robustness and sensitivity of BERT models predicting Alzheimer's disease from text is important for both developing better classification models and for understanding their capabilities and limitations. In this paper, we analyze how a controlled amount of desired and undesired text alterations impacts performance of BERT. We show that BERT is robust to natural linguistic variations in text. On the other hand, we show that BERT is not sensitive to removing clinically important information from text.