Programme:    B. Tech   Semester:     FIFTH  

Course:   ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING (COMMON TO ALL BRANCHES EXCEPT COMPUTER SCIENCE AND IT)

Artificial Intelligence and Machine Learning

Course Objectives:

● To learn about AI problems, techniques and their modelling as state space search, problem characteristics, Production System categories.

● To learn different uninformed and heuristic search strategies for solving AI problems with examples

● To learn theorem proving with predicate logic, resolution, rule-based inference with forward and backward chaining

● Inheritable knowledge representation using slot-filler structures and dealing with different forms of uncertain and implicit knowledge

● To introduce essential concepts of plan generation, Natural Language understanding and Expert Systems.

Course Outcomes:

● By the end of the course the student understands, applies, evaluates and creates AI solutions as they are

● Able to characterize and model AI problems in a state space search framework and identify appropriate production system category to solve them

● Able to understand and evaluate pros & cons of different heuristic search strategies and apply appropriate heuristic search for specific problem solving scenario.

● Able to represent domain knowledge in the form of predicates / rules and applies logic and inference for deducing the validity of a given assertion.

● Able to create problem specific slot-filler knowledge structures and apply statistical, fuzzy and non-monotonic reasoning methods aptly to solve real world problems involving any type of uncertainty.

● Able to understand basic concepts and approaches to natural language processing, plan generation and expert system development.

August 2024

Introduction to Artificial Intelligence:

Biological motivation for AI, Human Brain, Neural Network Representation, ANN architecture, Perceptron, Multi-Layer Perceptron structure, Back Propagation

September 2024

Machine Learning:

Introduction to Machine Learning, Different types of Machine Learning methods, supervised, semi supervised, unsupervised and reinforcement learning.

October 2024

Classification and Regression algorithms:

Difference between classification and Regression, Classification algorithms, KNN, SVM algorithms and its applications, Regression algorithms, Linear Regression, Decision Tree Regression and Random Forest Regression.

November 2024

Convolutional Neural Networks:

Introduction to convolutional neural networks, Basic principle, Architecture, Types of CNN Layer, Pooling Layers, Convolutional layers and Fully connected layers, Applications of CNN

November and December 2024

Advanced topics in Artificial Intelligence and Machine Learning:

DNN Model, Significance, Overview of DNN techniques and its applications, Generative Adversarial Network (GAN) Model, working principle of GAN and its applications

Text Books:

1.       Artificial Intelligence and Machine Learning by Vinod Chandra SS and Anand Hareendran S PHI Publications

2.       Artificial Intelligence – A modern Approaceh Stuart Russel and Peter Norvig

Reference books:

1.  Introduction to Artificial Intelligence by Ertel W (2018) Springer International Publishing

2.  Machine Learning and Artificial Intelligence by Joshi and Ameet V (2022) Springer International Publishing

Two mid examinations: One October 2nd week, another December 2nd Week. Best of these two will be considered 

Assignment one in September and the other in November.  Best of these two will be considered 

Model Question Paper 1:

Artificial Intelligence and Machine Learning

Time: Three hours                                Maximum: 70 Marks

Question No.1 is compulsory

Answer any FOUR questions from the remaining

All questions carry equal marks

  1. a)What is the perceptron model in artificial neural networks?[3]

b))What are the advantages and disadvantages of supervised learning?[3]

   c)What types of data are well-suited for SVM classification?[3]

   d)What are the main components of a typical CNN architecture?[3]

   e)Discuss the application of DNNs in image classification tasks.[2]

2 a)What are the differences between the human brain's neural networks and artificial neural networks?[7]

b)What is the basic architecture of an artificial neural network?[7]

3 a)What are the ethical considerations in developing and deploying machine learning systems?[7]

b)What is reinforcement learning, and how does it differ from other machine learning methods?[7]

4 a)What is the K-Nearest Neighbors (KNN) algorithm, and how does it work for classification tasks?[7]

  b)Describe the random forest regression algorithm and its advantages over individual decision trees.[7]

5 a)How do the depth and width of a CNN affect its performance and ability to learn complex features?[7]

b)How do filters (kernels) in convolutional layers learn to detect features in the input data?[7]

6 a)How does the adversarial training process in GANs help generate realistic samples?[7]

b)How do regularization methods, such as dropout and L1/L2 regularization, help prevent overfitting in DNNs?[7]

7 a)How does back propagation help in training neural networks?[7]

b)How do the depth and width of a CNN affect its performance and ability to learn complex features?[7]

Model Question Paper 2:

Artificial Intelligence and Machine Learning

Time: Three hours                                                                                Maximum: 70 Marks

Question No.1 is compulsory

Answer any FOUR questions from the remaining

All questions carry equal marks

  1. a)What are the main components of a neuron in the human brain?[3]

b)What are the challenges in semi-supervised learning?[3]

c)What is the goal of linear regression, and how does it find the best fit line?[3]

d)What is the fundamental principle behind the convolution operation used in CNNs?[3]

e)Why have DNNs become increasingly popular in recent years?[2]

2 a)How do the neural networks in the human brain inspire the design of artificial neural networks?[7]

b)How does the perceptron model differ from the multilayer perceptron model?[7]

3 a)What are the different types of machine learning methods?[7]

  b)What is machine learning, and how does it differ from traditional programming?[7]

4 a)Explain the concept of the decision boundary in classification problems.[7]

  b)What is the main difference between classification and regression problems in machine learning?[7]

5 a) Discuss the purpose of using activation functions in CNNs, and which functions are commonly used?[7]

b)What are the key historical developments that led to the popularity of CNNs in deep learning [7]                                                                   

6 a)What are the main components of a GAN, and how do they interact during the training process?[7]

b)What are some common techniques used in training DNN models, such as backpropagation and gradient descent?[7]

7 a)Compare the performance of linear regression, decision tree regression and random forest regression on datasets with different characteristics.[7]

b)What is the purpose of pooling layers in CNNs, and how do they contribute to the network's performance?[7]