Machine Learning (AL-405) - Important Questions

Explain the concepts of hypothesis space and inductive bias in machine learning and their importance in model learning.

Explain machine learning, its scope, limitations and real-world applications with examples.

Explain Principal Component Analysis (PCA) with mathematical steps and advantages, and compare PCA with Partial Least Squares (PLS).

Distinguish between supervised learning and unsupervised learning with examples.

Explain what Principal Component Analysis (PCA) is, with examples, and how it helps in dimensionality reduction of data.

Define Machine Learning and explain its perspectives and issues.

Define Machine Learning and explain how it differs from traditional programming with examples.

Explain the concept of Principal Component Analysis (PCA) and derive its mathematical formulation for dimensionality reduction and visualization.

Discuss how to evaluate the stability of a machine learning model. What metrics or methods can be used to assess how consistent a model's performance is across different datasets, environments, or perturbations? How does model stability impact its generalization ability?

Explain the need for inductive bias in decision tree learning.

Explain what a hypothesis space represents in machine learning and list its basic characteristics.

Discuss common limitations of machine learning.

Explain parallel processing / parallel distributed processing in neural networks and how it enables efficient computation, with diagram.

Explain the architecture of Multilayer Perceptron (MLP) and describe the backpropagation algorithm with diagram.

What are Vanishing Gradient and Exploding Gradient problems in detail?

Define backpropagation and explain its algorithm for computing gradients and updating weights in a neural network, with example/diagram.

What is a neural network and how is it represented mathematically, including the role of weights, biases, and activation functions?

What is a Multilayer Perceptron, its architecture, how it differs from a single-layer perceptron, and how hidden layers help solve complex problems?

Describe gradient descent and compare its variants SGD, Mini-batch Gradient Descent, Momentum, RMSProp, and Adam optimizers.

Explain the multi-layer perceptron model with a neat diagram.

Evaluate the effectiveness of biologically inspired neural network architectures in solving complex machine learning tasks.

What is the perceptron learning algorithm and how does it adjust connection weights to learn from data?

Discuss how the Multilayer Perceptron algorithm computes outputs from inputs through its interconnected layers.

How are neural network architectures represented graphically or conceptually in machine learning?

Explain the Random Forest algorithm and how it combines multiple decision trees for classification and regression.

Define decision tree and explain the working of decision tree / ID3 / CART learning algorithm with an example

Explain Naive Bayes classifier, including its fundamental principle based on Bayes theorem, probability calculation, assumptions, advantages, disadvantages and applications.

Compare linear regression and logistic regression with suitable examples.

Explain the difference between classification and regression models.

Explain linear regression in detail with example and list the assumptions to be met before using it.

Explain the concept of SVM for classification, the role of margin maximization, and the kernel trick with different types of kernels.

What is logistic regression for binary classification? Derive the logistic (sigmoid) function and explain cost function optimization during training.

Explain the bias-variance tradeoff in supervised learning and how decision trees and random forests handle bias and variance.

Explain the goal of SVM for classification of linearly separable data, including hyperplane, margin maximization and how to compute the margin.

Explain the concept of entropy in decision trees and how entropy affects construction and splitting of nodes

Discuss linear regression and derive the individual error and minimization (cost) functions.

Explain the Expectation-Maximization (EM) algorithm, including the E-step and M-step, why it is needed for missing/incomplete data, with an example such as training Gaussian Mixture Models.

Explain the k-means clustering algorithm in detail, including steps, objective function, how it partitions data, limitations and effect of choice of k

Explain Gaussian Mixture Model (GMM) for density estimation, including the Expectation-Maximization (EM) algorithm with flowchart/example.

Explain unsupervised learning and describe the k-means clustering algorithm with its advantages

Differentiate between agglomerative and divisive hierarchical clustering with examples.

Explain how adaptive hierarchical clustering works and how it adaptively determines the number of clusters from the data.

Use k-means clustering with initial centroids $m_1=2$ and $m_2=4$ and Euclidean distance to cluster the 1D data $\{2, 4, 10, 12, 3, 20, 30, 11, 25\}$ into two groups

Define unsupervised learning. Explain its significance and challenges. Discuss different approaches to clustering and dimensionality reduction, citing examples of common algorithms used.

Explain the difference between hard clustering and soft clustering, including algorithms, examples and use cases for each.

Explain how Genetic Algorithms are applied to optimize clustering, including chromosome representation, fitness function design, and genetic operations.

List applications of clustering and identify advantages and disadvantages of clustering algorithms.

Explain evolutionary optimization techniques in machine learning and give an overview of common evolutionary algorithms.

Explain resampling methods and cross-validation methods used in machine learning.

Explain hypothesis testing in machine learning with examples.

Explain measuring classifier performance using Accuracy, Precision, Recall and F1-Score.

Explain the key guidelines and fundamental steps for designing and conducting machine learning experiments, including reproducibility, dataset partitioning, and control of confounding variables for validity and generalizability.

Explain the importance and significance of comparing machine learning models over multiple datasets, including benefits, challenges, and robustness of insights

Explain the design and analysis of machine learning experiments in detail.

Explain the bias-variance trade-off in designing machine learning experiments, its relation to overfitting and underfitting, and how experimental design mitigates high bias or high variance.

Describe key elements of experimental design such as hypothesis formulation, variable manipulation, and control.

What is cross-validation in machine learning? Describe its purpose and how it helps in evaluating model performance.

Write short notes on hypothesis testing, subset selection, inductive bias and simulation.

Write short notes on cross validation, training and validation, cluster, and factors.

Write a short note on any two of the following:

Write a short note on any two of the following.