Essay

Explain why the two wake word error rates have different roles in the proposed performance goal.

Question: In a concise analytical response, explain how the optimizing-and-satisficing framework handles false negatives and false positives for the wake word detector.

Sample answer: Both error rates matter, but the proposed goal gives them different functions. The false positive rate becomes a satisficing metric: a detector is acceptable only if it produces no more than one false positive every 24 hours of operation. Among detectors that meet this requirement, the false negative rate is the optimizing metric, so the preferred detector is the one that minimizes missed wake words.

Key points:

  • False positives are wake-ups when no one said the wake word.
  • False negatives are failures to wake when someone says the wake word.
  • False negative rate is minimized as the optimizing metric.
  • False positives must not exceed one every 24 hours as the satisficing requirement.
  • Optimization occurs among detectors that satisfy the false-positive constraint.

Rubric: A strong response accurately defines both error types, identifies false negative rate as the optimizing metric, states the false-positive constraint, and explains how the constraint and optimization jointly determine the preferred detector.

0

1

Updated 2026-07-20

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Machine Learning Yearning @ DeepLearning.AI