Indian Institute of Technology Madras
BS Degree Programme · Examination & Study Paper
Machine Learning Techniques
Bias-variance, overfitting & validation
Examination Instructions & Guidelines
- Solve on notebook/paper first before checking solutions.
- MCQ: Exactly 1 option is correct. Negative marking applies.
- MSQ: One or more options correct. Select all to score.
- NAT: Numerical response. No negative marking.
- Scientific calculator allowed for all calculations.
- Official answer key is sealed at the end of this sheet.
Quiz 2 · 3 Aug 2025 (AN)Quiz 2
- Q1MCQ3 Marks·Choose one option
- ASetting 3, because it has the lowest training error, indicating the best fit to the training data.
- BSetting 2, because it has the lowest validation error and a relatively low training error, indicating a good generalization.
- CSetting 1, because it has the highest number of iterations and therefore likely to be more stable.
- DSetting 4, because it uses the highest learning rate and converges the fastest.
End Term · 13 Apr 2025 (AN)End Term
- Q2MSQ3 Marks·Choose all correct options
- AModel 1 will have a higher bias than Model 2.
- BModel 2 will have a lower variance than Model 1.
- CModel 1 is more likely to underfit.
- DModel 1 is more likely to overfit.
End Term · 22 Dec 2024 (FN)End Term
- Q3NAT5 Marks·Write the number
Your answer: - Q4MCQ1 Mark·Choose one option
Is the following statement true or false? For a fixed size of the training and the test set, increasing the complexity of the model always leads to an increment in the test error.
- ATRUE
- BFALSE
- Q5MSQ3 Marks·Choose all correct options
Consider a supervised ML problem where a model is trained using a training dataset, and its performance is evaluated on both the training and validation datasets. During training, the model’s loss is plotted over time for both the datasets. Assume that the loss L(t) is decreasing as the model learns. The following graph shows two curves, Curve (A) and Curve (B), representing the loss for the training and validation datasets, respectively:
- ACurve (A) represents the training loss.
- BCurve (B) represents the training loss.
- CCurve (A) represents the validation loss.
- DCurve (B) represents the validation loss.
- Q6MSQ3 Marks·Choose all correct options
Which of the following options are correct? Select all that apply.
- AUnderfitted models generally have less bias and high variance.
- BA decision tree with the maximum possible length may lead to overfitting.
- CWeak learners are those whose performance is slightly better than the random classifier.
- DAll the estimators in bagging can be trained parallelly.
End Term · 22 Dec 2024 (AN)End Term
- Q7MSQ3 Marks·Choose all correct options
- ACurve (A) represents the training loss.
- BCurve (B) represents the training loss.
- CCurve (A) represents the validation loss.
- DCurve (B) represents the validation loss.
- Q8NAT5 Marks·Write the number
Your answer: - Q9MCQ3 Marks·Choose one option
Which of the following options are not correct? Select all that apply.
- AUnderfitted models generally have less bias and high variance.
- BA decision tree with the maximum possible length may lead to overfitting.
- CWeak learners are those whose performance is slightly better than the random classifier.
- DAll the estimators in bagging can be trained parallelly.
- Q10MCQ1 Mark·Choose one option
Is the following statement true or false? For a fixed size of the training and the test set, increasing the complexity of the model always leads to an increment in the test error.
- ATrue
- BFalse