Huawei HCIE-AI Practice Questions - Huawei Certified ICT Expert - Artificial Intelligence (Written H13-541+ Lab + Interview) Exam

Access More Questions

Cross-entropy loss between a one-hot target y and softmax prediction p is H(y, p) = -Σ y_i log p_i. How does this relate to KL divergence?

Cross-entropy and KL divergence are unrelated quantities H(y, p) = KL(p || y) and is symmetric H(y, p) is mutual information between y and p H(y, p) = KL(y || p) + H(y), and since H(y) is constant for fixed y, minimizing cross-entropy equals minimizing KL(y || p)

Correct Answer:

H(y, p) = KL(y || p) + H(y), and since H(y) is constant for fixed y, minimizing cross-entropy equals minimizing KL(y || p)

Access more questions from this quiz

Continue to Huawei HCIE-AI Practice Questions - Huawei Certified ICT Expert - Artificial Intelligence (Written H13-541+ Lab + Interview) Exam for more practice questions and the full quiz experience.

Access More Questions