CNN Architecture Fundamentals 4 — Questions and Answers
Question 1: Which pooling operation is most commonly used in CNNs and what does it compute?
- Average pooling — computes the mean of values in the pooling window
- Max pooling — selects the maximum value in the pooling window (Correct answer)
- L2 pooling — computes the Euclidean norm of values in the window
- Stochastic pooling — randomly samples a value weighted by activation magnitude
Correct answer: Max pooling — selects the maximum value in the pooling window
Max pooling takes the largest activation within each region, retaining the strongest feature response and providing a degree of translation invariance.
Question 2: In DenseNet, how does each layer receive its input?
- Only from the immediately preceding layer
- From a learned weighted sum of all preceding layers
- From all preceding layers via concatenation (Correct answer)
- From alternating layers spaced two apart
Correct answer: From all preceding layers via concatenation
DenseNet concatenates feature maps from all preceding layers as input to each layer, maximizing feature reuse and enabling gradient flow through the entire network.
Question 3: What is the key innovation in MobileNetV2's inverted residual block compared to standard residual blocks?
- It expands channels before the depthwise convolution, then compresses with a linear bottleneck (Correct answer)
- It uses dilated convolutions to increase receptive field without more parameters
- It replaces ReLU with sigmoid to prevent dead neurons
- It applies group convolutions to split channels into independent subgroups
Correct answer: It expands channels before the depthwise convolution, then compresses with a linear bottleneck
MobileNetV2 inverts the bottleneck by expanding channels first, applying depthwise conv, then projecting back to a low-dimensional space with a linear activation.
Question 4: What does 'dilation' (atrous convolution) add to a standard convolution?
- Extra zero-padding around the input border
- Spaces between filter elements, enlarging the receptive field without increasing parameters (Correct answer)
- Additional convolutional channels at each layer
- A skip connection that bypasses the current layer
Correct answer: Spaces between filter elements, enlarging the receptive field without increasing parameters
Dilated convolution inserts gaps (holes) between filter weights, exponentially growing the receptive field while keeping the parameter count and resolution constant.
Question 5: In a fully convolutional network (FCN), what replaces the fully connected layers to enable pixel-wise predictions?
- Global max pooling across spatial dimensions
- Transposed convolutions (or bilinear upsampling) to recover spatial resolution (Correct answer)
- Additional pooling layers to collapse the feature map
- Dropout layers applied at every spatial location
Correct answer: Transposed convolutions (or bilinear upsampling) to recover spatial resolution
FCNs replace dense layers with convolutional ones and use transposed convolutions to upsample feature maps back to the input resolution for dense prediction.
Question 6: What is the purpose of the squeeze-and-excitation (SE) block in SENet?
- To reduce spatial resolution by 50% between convolutional blocks
- To recalibrate channel-wise feature responses by modeling inter-channel dependencies (Correct answer)
- To apply group normalization across subsets of channels
- To merge feature maps from different layers using element-wise addition
Correct answer: To recalibrate channel-wise feature responses by modeling inter-channel dependencies
The SE block globally pools feature maps, learns channel importance weights via two FC layers, and scales channels accordingly — acting as channel-wise attention.
Question 7: How does a transposed convolution (sometimes called deconvolution) differ from a standard convolution in terms of spatial output?
- It reduces spatial dimensions like a strided convolution
- It increases spatial dimensions, making it useful for upsampling (Correct answer)
- It maintains identical spatial dimensions using same padding
- It operates along the channel axis rather than the spatial axes
Correct answer: It increases spatial dimensions, making it useful for upsampling
Transposed convolution is the gradient operation of a forward convolution; it increases spatial size and is used in decoders, GANs, and segmentation networks.
CNN Architecture Fundamentals 4
Which pooling operation is most commonly used in CNNs and what does it compute?