Explain imbalance, metrics, bias-variance, Transformers vs. CNNs

Q: Explain imbalance, metrics, bias-variance, Transformers vs. CNNs

This question evaluates understanding of class imbalance strategies, the bias–variance tradeoff, selection and interpretation of evaluation metrics for highly imbalanced binary classification, and architectural trade-offs between Transformers and CNNs in the Machine Learning domain.

Q: How do I approach Machine Learning interview questions?

Machine Learning questions require understanding of core concepts and practice. PracHub provides solutions with explanations to help you master machine learning interviews.

Question

Fraud Detection (≈1% Positives): Imbalance Strategies, Bias–Variance, Metrics, and Model Choices

Context: You are building a binary classifier to flag fraudulent transactions, where only about 1% of transactions are fraud (positives). Answer the following:

Strategies for class imbalance
- Data level (e.g., resampling)
- Algorithm level (e.g., class weights, focal loss)
- Decision level (e.g., threshold tuning, cost-sensitive decisions)
Bias–variance tradeoff
- Define the tradeoff
- How to diagnose high bias vs. high variance in this scenario
- How to mitigate each
Evaluation metrics
- Select and justify metrics suitable for extreme imbalance: PR AUC vs. ROC AUC, precision@k, recall@k, F1/Fβ, calibration/Brier score
- Discuss when each is preferable
Transformers vs. CNNs
- Compare inductive biases, typical inputs, computational tradeoffs
- When to choose each for text, sequences, images, or tabular fraud features

Explain imbalance, metrics, bias-variance, Transformers vs. CNNs

Fraud Detection (≈1% Positives): Imbalance Strategies, Bias–Variance, Metrics, and Model Choices

Solution

Comments (0)

Explain imbalance, metrics, bias-variance, Transformers vs. CNNs

Overview

Fraud Detection (≈1% Positives): Imbalance Strategies, Bias–Variance, Metrics, and Model Choices

Solution

Comments (0)