Machine Learning and Neural Networks AIE Practice Questions (Page 10)

Questions 46–48

1. 46application · medium

   During training of a deep neural network, the loss on the training set decreases steadily, but the loss on the validation set begins to increase after a few epochs. Which action is most appropriate to address this problem?

   • AIncrease the learning rate so the model converges faster to a lower training loss
   • BStop training early when validation loss starts to rise, or apply regularization such as dropout
   • CRemove the activation functions from the hidden layers to simplify the model
   • DIncrease the number of training epochs to allow the model to fit the training data even better
   Reveal Solution Discussion Select an answer first
2. 47application · medium

   A data scientist is building a neural network to predict house prices from features like square footage and number of bedrooms. The output layer must produce a continuous value. Which activation function is most appropriate for the output neuron?

   • ASoftmax activation function
   • BSigmoid activation function
   • CLinear (or identity) activation function
   • DReLU activation function
   Reveal Solution Discussion Select an answer first
3. 48expert · medium

   A company has a rule-based system for detecting fraudulent transactions that uses manually defined thresholds. The system misses many new fraud patterns. They want to switch to a machine learning approach. What is the primary advantage of using machine learning over the current rule-based system?

   • AMachine learning does not require any historical data to train
   • BMachine learning eliminates the need for human oversight entirely
   • CMachine learning is guaranteed to catch all fraudulent transactions
   • DMachine learning can automatically learn complex patterns from data and adapt to new fraud patterns without manual rule updates
   Reveal Solution Discussion Select an answer first