Scribd
Upload
31 views3 pages

ML Theory Questions Final

Uploaded by

Aniket Kumar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
31 views3 pages

ML Theory Questions Final

Uploaded by

Aniket Kumar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Machine Learning Theory Questions (Without Derivations)

1. Foundations of AI/ML/DS
What is the difference between Artificial Intelligence (AI), Machine Learning (ML), and Data
Science (DS)?

Compare supervised, unsupervised, and reinforcement learning with examples.

What is the difference between regression and classification problems?

Explain the concept of overfitting and underfitting with an example.

What is the bias-variance trade-off in model learning?

2. Linear Models
What is linear regression? What kind of problems can it solve?

What is the cost function used in linear regression? Why is MSE commonly used?

Explain the difference between R², MSE, MAE, and SSE.

3. Regularization
What is regularization in ML? Why is it needed?

Explain L1 (Lasso), L2 (Ridge), and ElasticNet regularization.

How does regularization help in controlling overfitting?

4. Classification and Metrics


What is logistic regression? How does it model probabilities?

How is logistic regression extended to multi-class classification?

What is softmax regression? How is it different from binary logistic regression?

What is one-hot encoding? Why is it used in classification problems?

List and explain classification metrics: Accuracy, Precision, Recall, F1 Score, Specificity,
Confusion Matrix.

What are commonly used loss functions for logistic and softmax regression?

5. Probabilistic Models
What is maximum likelihood estimation (MLE)? How is it used in model training?

What is a normal distribution? What does it represent?


Describe the multivariate Gaussian distribution. How does the covariance matrix affect its
shape?

What is Naive Bayes classifier? What assumption does it make?

Compare LDA and QDA in terms of assumptions and application.

6. Decision Trees and Ensembles


How do decision trees work for classification and regression?

What is Gini impurity? How is it used in decision trees?

What is the difference between bagging and boosting?

What is a random forest? How does it reduce variance?

Name and briefly explain types of boosting algorithms: AdaBoost, Gradient Boosting,
XGBoost, LightGBM.

7. Instance-Based and Clustering


How does K-Nearest Neighbors (KNN) work? What are its advantages and limitations?

Explain the K-means clustering algorithm step by step.

What is the difference between clustering and classification?

8. Dimensionality Reduction
What is the objective of Principal Component Analysis (PCA)?

What is the role of the covariance matrix and eigenvectors in PCA?

Compare PCA and LDA in terms of use-case and assumptions.

What is the difference between feature selection and dimensionality reduction?

9. Neural Networks
Describe the architecture of a basic feedforward neural network.

What is the universal approximation theorem?

Explain activation functions. Why are they important?

List and compare types of activation functions: Sigmoid, Tanh, ReLU, Leaky ReLU, Softmax.

What is gradient descent? How does it update parameters?

What is backpropagation? How is it related to learning in neural networks?


10. Physics-Informed Neural Networks (PINNs)
What is a Physics-Informed Neural Network?

How does a PINN differ from a traditional neural network?

How is the loss function constructed in a PINN?

Where are PINNs useful in engineering and physical sciences?

What are the advantages of PINNs over classical numerical methods?

11. General Conceptual & Applied


How do you choose the right ML model for a given problem?

What are the assumptions made by linear and logistic regression?

What are some challenges in training deep neural networks?

What is the role of learning rate in training a neural network?

What does convergence mean in the context of gradient descent?

Name some popular optimizers other than basic gradient descent. What are their
advantages?

What are some hyperparameters in neural networks and how can you tune them?

You might also like