Case context: You are training a traffic-sign classifier. During dev-set error analysis, you find a photo that clearly shows a stop sign. Your model predicts "stop sign" correctly, but the example is counted as an error because the dev-set label says "yield sign".

Question: What is the data issue here? Name the type of problem, explain who introduced it, and identify which part of the labeled pair (x, y) is wrong.

Sample answer: This is a mislabeled example. The mistake was introduced by a human annotator before the model saw the data. In the pair (x, y), the input x is the traffic-sign image, and the incorrect part is the label y, which should have been "stop sign" instead of "yield sign".

Key points:

• The issue is a mislabeled example.
• The error came from a human labeler before model training or evaluation.
• The incorrect component of (x, y) is the label y.

Rubric: The response must identify the example as mislabeled. It must state that a human annotator introduced the error before the algorithm saw the data. It must also specify that y, not x, is the incorrect part of the pair.