Skip to main content
Fig. 4 | AI Perspectives & Advances

Fig. 4

From: Corner cases in machine learning processes

Fig. 4

The four diagrams illustrate our definition of corner cases using a toy example with two classes (red and blue). (a) Prediction probability. (b) Task-specifc weighting. (c) Minimal expected loss (no weighting). (d) Minimal expected relevancy-weighted loss (without threshold), see Eq. 8

Back to article page