H13-311_V3.5 · Question #69
In neural networks, weights are often shareD. Which of the following neural networks will share weights?
The correct answer is B. Convolutional Neural Network D. Recurrent neural network. Convolutional Neural Networks (CNNs) share weights through their filters, where the same filter (kernel) is applied across different spatial locations of the input, meaning all neurons in a feature map share identical weights. Recurrent Neural Networks (RNNs) share weights…
Question
In neural networks, weights are often shareD. Which of the following neural networks will share weights?
Options
- Asensor
- BConvolutional Neural Network
- CFully connected neural network
- DRecurrent neural network
How the community answered
(21 responses)- A10% (2)
- B76% (16)
- C14% (3)
Explanation
Convolutional Neural Networks (CNNs) share weights through their filters, where the same filter (kernel) is applied across different spatial locations of the input, meaning all neurons in a feature map share identical weights. Recurrent Neural Networks (RNNs) share weights across time steps, applying the same set of weights at each step when processing sequential data. A "sensor" (option A) is not a type of neural network at all - it is an input device that collects data, so it has no weights to share. A fully connected network (option C) does not share weights; every connection between neurons has its own independent weight, making it the most parameter-heavy architecture.
Memory tip: Think "reuse" - CNNs reuse a filter across space, RNNs reuse weights across time. Both repeat their weights, unlike fully connected networks where every weight is unique.
Topics
Community Discussion
No community discussion yet for this question.