Home › Claude › aifoundations › alexnet_cnn_imagenet › Removing any single middle convolutional layer o…
Removing any single middle convolutional layer of AlexNet costs roughly
ANo measurable drop in accuracy
BA 2% drop in top-1 accuracy
CA 10% drop in top-1 accuracy
DA 20% drop in top-1 accuracy
Answer & Solution
Correct answer: B. A 2% drop in top-1 accuracy
About 2% loss in top-1 if any middle layer is removed. The authors take this as evidence that depth is essential — not just total parameter count.
Related questions
Image similarity in AlexNet's last hidden layer is measured byThe SGD update rule for AlexNet usesAlexNet was trained for roughly how many epochs on ImageNet?At test time dropout is approximated byAlexNet uses overlapping pooling withAlexNet's second form of data augmentation isAlexNet's first form of data augmentation isWhy are ReLU networks faster to train than tanh networks?