Selected MCQ

Current Question
What role does the "loss function" serve in training a neural network?
  • A. Generate predictions
  • B. Aggregate features
  • C. Calculate error to guide learning
  • D. Normalize data
Correct Answer: C
Explanation:
The loss function quantifies the difference between predictions and true values, guiding weight adjustments.
Related Question 1
What is the main function of dropout layers in a neural network?
  • A. Speed up training by skipping layers
  • B. Prevent co-adaptation of neurons to reduce overfitting
  • C. Ensure output is always positive
  • D. Increase network depth
Correct Answer: B
Explanation:
Dropout randomly turns off neurons during training, reducing overfitting.
Related Question 2
What is the primary advantage of using a GPU over a CPU for training deep neural networks?
  • A. Faster single-thread performance
  • B. Large memory capacity
  • C. Parallel processing of many operations
  • D. Built-in neural network instructions
Correct Answer: C
Explanation:
GPUs can perform many operations in parallel, greatly speeding up matrix computations in neural network training.
Related Question 3
Which activation function is commonly used in hidden layers of deep neural networks due to its ability to mitigate vanishing gradient problems?
  • A. Sigmoid
  • B. Tanh
  • C. ReLU
  • D. Softmax
Correct Answer: C
Explanation:
ReLU (Rectified Linear Unit) is popular for hidden layers because it accelerates convergence by avoiding vanishing gradients.
Related Question 4
Which term describes training a model on multiple GPUs or servers simultaneously?
  • A. Distributed training
  • B. Batch processing
  • C. Layer-wise training
  • D. Fusion
Correct Answer: A
Explanation:
Distributed training spreads the training process across multiple devices or machines.
Related Question 5
Which loss function is typically used for binary classification?
  • A. Softmax loss
  • B. Mean squared error
  • C. Cross-entropy loss
  • D. Huber loss
Correct Answer: C
Explanation:
Cross-entropy loss (binary or logistic) is commonly used for binary classification tasks.