FRM Part I · FRM Exam Part I · Machine Learning and Prediction
Compared with a logistic regression using the same inputs, a deep neural network used for a bank's default prediction is most likely to present which challenge?
The main challenge is lower interpretability. A deep network spreads the input-output relationship across many nonlinear weights, making it hard to explain predictions to regulators and risk governance. It can model nonlinearities, accepts continuous inputs, and does not guarantee a global minimum.
- ALower interpretability, since the relationship between inputs and output is spread across many nonlinear weightsCorrect
- BInability to model any nonlinear relationship between inputs and default
- CA requirement that all inputs be binary variables
- DGuaranteed convergence to the global minimum of the loss function
Explanation
Deep networks can capture nonlinearities and interactions but are hard to interpret, which matters for model governance and regulation. They handle nonlinearity well, accept continuous inputs, and gradient training generally does not guarantee a global minimum.
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.
More Machine Learning and Prediction questions
- A node in a classification tree holds 100 observations: 50 defaults and 50 non-defaults. A candidate split sends 40 observations left (35 de…
- In a K-means run with K = 2, five one-dimensional observations are 1, 2, 4, 9 and 10. The initial centroids are 2 and 9. After assigning eac…
- Which statement correctly distinguishes a random forest from plain bagging of decision trees?
- A neural network hidden node receives inputs x1 = 2 and x2 = -1. The weights are w1 = 0.5 and w2 = 1.5, and the bias is 0.25. The node uses …
- In k-fold cross-validation with k = 5 applied to a sample of 500 observations, how many observations are used to train the model in each ite…
- A risk analyst applies principal components analysis (PCA) to a set of 10 highly correlated yield-curve variables. Which statement best desc…