Figure 3:

Accuracy of our model (Multi-Risk) for both tasks with comparison to two baseline methods. Showing the median, 25th, and 75th percentiles over 10-fold cross validation.