Machine Learning 2025

(a) Gradient Descent is an example of:

1. Evolutionary algorithm
2. Sampling algorithm
3. Optimization algorithm
4. Clustering algorithm

(b) Which regression technique performs feature selection?

1. Ridge Regression
2. Linear Regression
3. LASSO
4. Polynomial Regression

(c) PCA chooses components that:

1. Maximize variance
2. Maximize covariance
3. Minimize variance
4. Increase dimensionality

(d) In Bayesian Learning, a conjugate prior is chosen because it:

1. Simplifies posterior calculation
2. Eliminates noise
3. Improves accuracy
4. Selects features

(e) Logistic Regression uses:

1. Linear activation
2. ReLU activation
3. Sigmoid activation
4. Hyperbolic tangent

(f) Perceptron works only on:

1. Linear separable data
2. Non-linear data
3. High-dimensional data
4. Unsupervised data

(g) Bagging mainly helps in reducing:

1. Bias
2. Variance
3. Noise
4. Over-sampling

(h) K-medoids is more robust to

1. Large data
2. Gaussian distribution
3. Outliers
4. High variance

(i) Bayesian networks represent:

1. Undirected dependencies
2. Fully connected graph
3. Directed acyclic graph
4. Weighted graph

(j) K-means uses:

1. Manhattan distance
2. Medoids
3. Centriods
4. Probabilities

(a) Define hypothesis space nd inductive bias with examples.

(b) Explain the difference between supervised, unsupervised and reinforcement learning with real world examples.

(a) What is bias-variance trade-off? Describe K-fold cross-validation and its importance.

(b) Draw the perceptron network with the notation. Derive an equation of gradient descent rule to minimize the error.

(a) Explain PCA with derivation of eigen decomposition and dimensionality reduction.

(b) Compare ML, MAP and Bayes estimators using an example.

(a) Describe ANN architecture and derive back propagation equations.

(b) Explain Decision Tree learning using Gini Index & Information Gain.

(a) Explain ensemble learning. How do Bagging and Boosting reduce error?

(b) Discuss hypothesis testing and its role in ML model comparison.

(a) Compare K-means, K-medoids, and spectral clustering in detail.

(b) Derive the Expectation-Maximization algorithm step-by-step.

(a) Explain reinforcement learning with policy, reward, and value functions.

(b) Describe Bayesian Networks and inference techniques.

(a) Support Vector Machine

(b) Linear Regression

(c) Gaussian Mixture Model

(d) AdaBoost algorithm