Question 5
UnclassifiedConvolutional Neural Networks excel at image tasks primarily because of:
Correct answer: B
Explanation
Convolutional Neural Networks use "local connectivity" and "weight-sharing" through convolutional filters, which lets them detect patterns like edges and textures anywhere in an image. This exploits the image’s spatial structure, reducing the number of parameters and making the model efficient for image tasks.
Why each option is right or wrong
A. Fully connected layers between every pixel pair
B. Local connectivity and weight-sharing via convolutional filters that exploit spatial structure
Convolutional neural networks are built around convolutional layers, where each filter is applied to small receptive fields rather than the full input, so nearby pixels are processed together and the same filter weights are reused across the entire image. That architectural choice preserves spatial relationships and makes the model sensitive to patterns such as edges, corners, and textures regardless of where they appear, while also cutting the parameter count dramatically compared with fully connected layers.
C. Random projections
D. Pre-trained word embeddings