Neural networks · Question 17 of 60

Why are special coverage measures needed for neural networks?

1 answer is correct. Easy

1. A Because neural networks may not have source code, so classic measures are legally inadmissible
2. B Because statement coverage always yields 0% for neural networks, no matter how many tests run
3. C Because their behavior is not determined by coded paths but by learned weights, bias values and activations
4. D Because neural networks could otherwise not be tested at all and coverage is the only test