Case Study

Diagnose a pipeline error where a Siamese cat is misclassified despite a correct bounding box from the detector.

Case context: You are developing a two-stage pipeline to detect and classify cat breeds. For a test image of a Siamese cat, the system produces a wrong breed classification. Upon visual inspection, you notice that the cat detector has successfully outputted an appropriate bounding box around the cat.

Question: Based on this inspection, which component in the pipeline is at fault for the wrong Siamese-cat decision, and how do you attribute this error?

Sample answer: The error should be attributed to the cat breed classifier. Since the cat detector outputted an appropriate bounding box, it successfully performed its job of locating the cat. The failure to correctly label the Siamese cat must therefore be diagnosed as a failure of the cat breed classifier component.

Key points:

  • Identify the cat breed classifier as the component at fault.
  • Establish that the cat detector successfully completed its task by outputting an appropriate bounding box.
  • Attribute the incorrect Siamese-cat decision to the breed classifier stage.

Rubric: The learner must diagnose that the cat breed classifier is at fault. The response must justify this by stating that the cat detector successfully performed its job by producing an appropriate bounding box.

0

1

Updated 2026-05-26

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Related