Skip to content

FRM Part I · FRM Exam Part I · Machine Learning and Prediction

A risk analyst builds a feedforward neural network to predict loan defaults. Which description best characterizes the role of a hidden layer in this network?

A hidden layer applies weights and a nonlinear activation function to its inputs, generating intermediate features that subsequent layers use. This is what lets neural networks capture nonlinear relationships. It is unrelated to averaging trees, outlier removal, or setting the learning rate.

  1. AIt applies weights and a nonlinear activation function to the inputs, creating intermediate features used by later layersCorrect
  2. BIt computes the final default probability by averaging the outputs of independent decision trees
  3. CIt removes all observations whose input values lie outside two standard deviations of the mean
  4. DIt fixes the learning rate to a constant value so that gradient descent converges in one step

Explanation

Hidden layers take weighted sums of inputs and pass them through nonlinear activation functions, producing intermediate representations that later layers combine. Averaging independent trees describes a random forest, not a hidden layer. Hidden layers do not clean data or set the learning rate.

Did you get it right without looking?

One question tells you little. A timed set on Machine Learning and Prediction shows your real accuracy, how long you take and where you lose marks.

Practise Machine Learning and PredictionFree diagnostic mock

More Machine Learning and Prediction questions