Performance measures for image classification
4 important questions on Performance measures for image classification
What is a single-value performance measure?
Name three single-value measures and how they work
- F-score: Harmonic mean of precision and recall (2PR)/(P+R)
- Receiver-Operating Characteristic (ROC), R vs P (x vs y) (curve based)
- Fix precision or recall at a certain value (curve based)
What is the accuracy in classification?
- Higher grades + faster learning
- Never study anything twice
- 100% sure, 100% understanding
How is the accuracy measure short-sided in describing the quality of a binary classifier? Give two
- Skewness in class distribution causes the accuracy to describe dominantly about the most prevalent (majority) class(es)
- Some misclassification may be worse than others, yet cannot be capsulated in this measure.
The question on the page originate from the summary of the following study material:
- A unique study and practice tool
- Never study anything twice again
- Get the grades you hope for
- 100% sure, 100% understanding