Course Objective
The course aims to introduce the fundamental principles of intelligent systems to students. This involves ideas about the characteristics of intelligent systems, knowledge representation schemes, logic and inference mechanisms. The course helps the learner to understand the design of self-learning systems along with some of their typical applications in the emerging scenario where the business world is being transformed by the progress made in machine learning.
Course Outcomes
Students will be able to:
CO1: Explain the fundamental concepts of intelligent systems and their architecture.
CO2: Illustrate uninformed and informed search techniques for problem solving in intelligent systems.
CO3: Solve Constraint Satisfaction Problems using search techniques.
CO4: Represent AI domain knowledge using logic systems and use inference techniques for reasoning in intelligent systems.
CO5: Illustrate different types of learning techniques used in intelligent systems.
Syllabus
Module I (Introduction)
Introduction – What is Artificial Intelligence(AI) ? The Foundations of AI, History of AI, Applications of AI. Intelligent Agents – Agents and Environments, Good behavior: The concept of rationality, nature of Environments, Structure of Agents.
Module II (Problem Solving)
Solving Problems by searching-Problem solving Agents, Example problems, Searching for solutions, Uninformed search strategies, Informed search strategies, Heuristic functions.
Module III (Search in Complex environments)
Adversarial search – Games, Optimal decisions in games, The Minimax algorithm, Alpha-Beta pruning. Constraint Satisfaction Problems – Defining CSP, Constraint Propagation- inference in CSPs, Backtracking search for CSPs, Structure of CSP problems.
Module IV (Knowledge Representation and Reasoning)
Logical Agents – Knowledge based agents, Logic, Propositional Logic, Propositional Theorem proving, Agents based on Propositional Logic. First Order Predicate Logic – Syntax and Semantics of First Order Logic, Using First Order Logic, Knowledge representation in First Order Logic. Inference in First Order Logic – Propositional Vs First Order inference, Unification and Lifting, Forward chaining, Backward chaining, Resolution.
Module V (Machine Learning)
Learning from Examples – Forms of Learning, Supervised Learning, Learning Decision Trees,
Evaluating and choosing the best hypothesis, Regression and classification with Linear models.
Text Books
1. Stuart Russell and Peter Norvig. Artificial Intelligence: A Modern Approach, 3rd Edition, Prentice Hall.
Reference Books
1. Nilsson N.J., Artificial Intelligence – A New Synthesis, Harcourt Asia Pvt. Ltd.
Curriculum
- 5 Sections
- 20 Lessons
- 12 Weeks
- Module 01Artificial Intelligence (AI) is a branch of computer science that aims to create systems capable of performing tasks that typically require human intelligence, such as learning, reasoning, and problem-solving. The foundations of AI lie in disciplines like mathematics, logic, neuroscience, computer engineering, and cognitive science. The history of AI began in the 1950s, evolving through symbolic AI, expert systems, machine learning, and modern deep learning approaches. AI has numerous applications, including speech recognition, autonomous vehicles, healthcare diagnostics, and recommendation systems. Intelligent agents are core to AI, functioning within environments and making rational decisions based on perception, where their behavior is influenced by the environment’s nature and their internal structure.3
- Module 02Solving Problems by Searching is a foundational concept in Artificial Intelligence where intelligent agents find solutions by exploring possible actions. Problem-solving agents are designed to decide what to do by identifying goals and searching for sequences of actions to achieve them. This approach is applied to many example problems such as pathfinding, puzzle-solving, and game playing. The process involves searching for solutions through various strategies, broadly categorized into uninformed search strategies (like BFS and DFS, which have no additional information) and informed search strategies (like A*, which use problem-specific knowledge). Heuristic functions guide informed search by estimating the cost from a given state to the goal, improving efficiency and effectiveness.4
- Module 03Adversarial search is used in game-playing scenarios where multiple agents (players) compete with opposing goals, requiring intelligent decision-making. In such games, optimal decisions aim to maximize a player's advantage while minimizing the opponent's, typically modeled using the Minimax algorithm, which assumes both players play optimally. Alpha-Beta pruning enhances Minimax by eliminating branches that won’t affect the final decision, improving efficiency without compromising accuracy. Constraint Satisfaction Problems (CSPs) involve finding values for variables within specific constraints, such as in puzzles like Sudoku or scheduling problems. CSPs are solved using constraint propagation (to infer variable domains), backtracking search (to explore possible assignments), and by exploiting the problem’s structure (like tree-based or graph-based representations) to improve performance.4
- Module 04Logical agents are intelligent systems that make decisions based on a knowledge base using formal logic. Knowledge-based agents use propositional and first-order logic to represent facts about the world and apply inference rules to derive new knowledge. Propositional logic deals with simple true/false statements, and agents based on it use theorem proving methods to reason. First Order Predicate Logic (FOPL) extends propositional logic by introducing quantifiers and predicates, allowing more expressive knowledge representation. Inference in FOPL includes advanced techniques like unification, forward and backward chaining, and resolution, which enable deeper reasoning compared to propositional logic.5
- Module 05Learning from examples is a core concept in machine learning where systems improve their performance by analyzing data. Forms of learning include supervised, unsupervised, semi-supervised, and reinforcement learning, depending on the availability of labeled data. In supervised learning, the model learns from input-output pairs to make future predictions. One popular supervised learning method is decision trees, which recursively split data based on feature values to classify or predict outcomes. Evaluating and selecting the best hypothesis involves metrics like accuracy, precision, and error rate, while regression and classification with linear models help in predicting continuous values or assigning class labels using linear relationships.4
Target audiences
- Undergraduate students of computer science and its disciplines