1 Introduction
2 Overview of the Related Work
2.1 How Does MC-Dropout Work?
2.2 Predictive Uncertainty Using MC-Dropout
2.3 Uncertainty and Risk
3 Experiments and Discussions
3.1 Classification Using Synthetic Data: Toy Example
3.2 Classification Using Real Data: Elliptic Data
Model | Accuracy | \( F_{1} \)-score | AUC |
---|---|---|---|
Standard (no dropout) | 0.943 | 0.619 | 0.892 |
MC-dropout | 0.966 | 0.729 | 0.894 |