Dataset and label checks · Question 60 of 60

A team has a new labeled dataset of traffic scenes and knows a comparable public dataset. For "pedestrian", the label share is five times as high as in the comparison dataset. Which approach is this, and what follows from it?

1 answer is correct. Hard

1. A Dataset constraint testing; the type rule for labels is violated
2. B Model loss analysis; the model must be retrained because of high loss
3. C Data distribution analysis; the deviation may indicate anomalies and incorrect labels
4. D Disparate impact analysis; the model is biased against pedestrians