CNN Classic CNN Architectures 4 — Questions and Answers
Question 1: Which classic CNN architecture introduced Local Response Normalization (LRN)?
- LeNet-5
- VGGNet
- AlexNet (Correct answer)
- GoogLeNet
Correct answer: AlexNet
AlexNet introduced Local Response Normalization, which normalizes across adjacent feature maps to encourage competition between neurons — though later research found it rarely beneficial.
Question 2: How many parameters does GoogLeNet have compared to AlexNet?
- GoogLeNet has roughly 12x more parameters
- GoogLeNet has roughly the same number of parameters
- GoogLeNet has roughly 12x fewer parameters (Correct answer)
- GoogLeNet has roughly 4x fewer parameters
Correct answer: GoogLeNet has roughly 12x fewer parameters
AlexNet has ~60M parameters while GoogLeNet has only ~5M, making GoogLeNet roughly 12x more parameter-efficient due to its Inception modules.
Question 3: What pooling strategy does GoogLeNet use at the end of the network instead of fully connected layers?
- Max pooling
- Average pooling (Correct answer)
- Fractional max pooling
- Adaptive average pooling
Correct answer: Average pooling
GoogLeNet uses global average pooling before the final classifier, drastically reducing parameters compared to flattening into large fully connected layers.
Question 4: What is the approximate top-5 error rate AlexNet achieved on ImageNet ILSVRC 2012?
- 25.8%
- 15.3% (Correct answer)
- 10.9%
- 5.0%
Correct answer: 15.3%
AlexNet achieved a top-5 error of 15.3%, which was dramatically better than the runner-up's 26.2%, demonstrating deep CNN superiority.
Question 5: Which VGGNet variant performs best on ImageNet while balancing accuracy and parameter count?
- VGG-11
- VGG-13
- VGG-16 (Correct answer)
- VGG-19
Correct answer: VGG-16
VGG-16 is the most commonly used variant, offering a strong balance between accuracy and parameter efficiency compared to VGG-19 with only marginal gains.
Question 6: In what year did ResNet win the ImageNet Large Scale Visual Recognition Challenge (ILSVRC)?
- 2012
- 2013
- 2014
- 2015 (Correct answer)
Correct answer: 2015
ResNet won ILSVRC 2015 with a top-5 error of 3.57%, surpassing human-level performance on the ImageNet classification task.
Question 7: What is the input image size expected by AlexNet?
- 224x224
- 227x227 (Correct answer)
- 256x256
- 299x299
Correct answer: 227x227
AlexNet accepts 227x227 pixel RGB images as input (sometimes cited as 224x224 due to an error in the original paper, but the actual implementation uses 227x227).
CNN Classic CNN Architectures 4
Which classic CNN architecture introduced Local Response Normalization (LRN)?