Skip to main content
. 2023 Mar 13;25:e35568. doi: 10.2196/35568

Figure 5.

Figure 5

Our feature-rich modeling approach to automated quality assessment. The actual form of the output layer depends on the task. The schematic shows all possible task-specific heads, but we train and evaluate the model for each task (T1 [b], T1 [a], T2 [a], and T2 [b]) independently. FFNN: feed-forward neural network. NUM: numerical. CAT: categorical.