Rethinking Generalization: The Error Rate
Beginner
The problem, from first principles
Generalization experiments measure errors on the training labels and on held-out labels. The error rate is the simplest measure: the fraction of predictions that disagree with the label.
From theory to code
Implement error_rate(pred, y), returning the fraction of mismatched predictions.
Constraints
- Returns a Python float.
Hints
Hint 1
Compare the arrays elementwise, average the boolean mismatches.
Keywords
research-papersclassical-mllearning-theorygeneralization
Python
Run your code to see results
No results yet
Run your code to check it against the examples, or submit to run every test.