Q1. Multivariate Normal Distribution
A multivariate Gaussian is fully defined by:
A. Mean only
B. Covariance only
C. Mean and covariance
D. Variance only
✅ Answer: C
Q2. Covariance Matrix Property
The covariance matrix Σ must be:
A. Negative
B. Diagonal only
C. Symmetric
D. Random
✅ Answer: C
Q3. Gaussian Process Definition
A Gaussian Process is:
A. Parametric model
B. Deterministic function
C. Distribution over functions
D. Classification model
✅ Answer: C
Q4. GP Output
Gaussian Processes provide:
A. Only prediction
B. Only variance
C. Prediction + uncertainty
D. Only labels
✅ Answer: C
Q5. Kernel Function Role
The kernel defines:
A. Mean
B. Covariance structure
C. Labels
D. Loss function
✅ Answer: B
Q6. Squared Exponential Kernel
If x₁ = x₂, covariance is:
A. 0
B. 1
C. ∞
D. −1
✅ Answer: B
Q7. Distance Effect on Covariance
If |x₁ − x₂| increases:
A. Covariance increases
B. Covariance decreases
C. No change
D. Becomes negative
✅ Answer: B
Q8. GP Length Scale (λ)
Large λ leads to:
A. Wiggly function
B. Smooth function
C. Random output
D. No prediction
✅ Answer: B
Q9. Signal Variance (σ²f)
Increasing signal variance causes:
A. Smaller outputs
B. Larger variation
C. No effect
D. More noise
✅ Answer: B
Q10. Noise Variance (σ²n)
Noise variance controls:
A. Smoothness
B. Data noise level
C. Distance
D. Kernel type
✅ Answer: B
Q11. GP Model Characterisation
A GP is fully defined by:
A. Kernel only
B. Mean only
C. Mean + covariance
D. Hyperparameters only
✅ Answer: C
Q12. One-vs-All Classification
For k classes, number of classifiers:
A. 1
B. k
C. k²
D. k(k−1)/2
✅ Answer: B
Q13. One-vs-One Classification
Number of classifiers:
A. k
B. k−1
C. k(k−1)/2
D. 2k
✅ Answer: C
Q14. One-Hot Encoding
A categorical variable with c values becomes:
A. 1 variable
B. c variables
C. c² variables
D. log(c) variables
✅ Answer: B
Q15. Imbalanced Data Issue
Models tend to:
A. Underpredict majority class
B. Overpredict majority class
C. Ignore data
D. Always balance
✅ Answer: B
Q16. Undersampling
Undersampling means:
A. Adding data
B. Removing majority samples
C. Increasing noise
D. Feature selection
✅ Answer: B
Q17. Reproducibility Definition
Reproducibility means:
A. Same author repeats results
B. Others reproduce results with own implementation
C. Same dataset only
D. Same code only
✅ Answer: B
Q18. Repeatability
Repeatability means:
A. Others reproduce results
B. Same author repeats experiment
C. Different datasets
D. Random runs
✅ Answer: B
Q19. Random Seed
Using same seed ensures:
A. Different results
B. Same results
C. Faster training
D. Better accuracy
✅ Answer: B
Q20. McNemar’s Test
Used for:
A. Clustering
B. Regression
C. Comparing classifiers
D. Feature selection
✅ Answer: C
Top comments (0)