Here are the top Neural Networks interview questions to prepare for your next role.
1️⃣ Do you need to normalize inputs before training in neural networks?
- A) Yes, because it ensures faster convergence
- B) No, it doesn't affect the training process
- C) Yes, because it avoids potential numerical issues
- D) No, normalization is only required in image processing tasks
2️⃣ How does bias function within a neural network?
- A) It adjusts the output along with the weighted input to improve model's predictions.
- B) It scales all input values by a fixed factor before passing them through the activation function.
- C) It reduces the dimensionality of input data prior to processing.
- D) It prevents overfitting by adding random noise to input values during training.
3️⃣ In which layer is the ReLU activation function usually used?
- A) Input Layer
- B) Hidden Layer
- C) Output Layer
- D) Normalization Layer
4️⃣ What is the vanishing gradient problem?
- A) When gradients become too small to update the network effectively during training.
- B) When gradients become too large causing the network weights to change erratically during training.
- C) When gradients stabilize and do not change much after a few epochs of training.
- D) When gradients oscillate between positive and negative values leading to slow convergence.
5️⃣ Assume that dropout rate is equal to 0.7. What does it mean?
- A) 70% of the neurons are dropped during training.
- B) 70% of the neurons are retained during training.
- C) 70% of the neurons are dropped during inference.
- D) 70% of the neurons are retained during inference.