MODULE 1: Problem Solving and Scope of AI
Introduction to Artificial Intelligence:
AI is the simulation of human intelligence in machines programmed to think and learn.
AI Applications:
Games: Pathfinding in chess, tic-tac-toe, etc.
Theorem Proving: Solving complex proofs using algorithms.
Natural Language Processing (NLP): AI for text analysis, machine translation, speech recognition.
Vision and Speech Processing: Image recognition, voice assistants (e.g., Siri, Alexa).
Robotics: Industrial automation, autonomous vehicles, humanoid robots.
Expert Systems: AI systems solving domain-specific problems (e.g., medical diagnosis).
Solving Problems Using Search:
Uninformed Search: Blind search strategies without prior knowledge.
Examples: Breadth-First Search (BFS), Depth-First Search (DFS), Uniform-Cost Search.
Informed Search: Uses heuristics to improve search efficiency.
Examples: Greedy Best-First Search, A* Algorithm.
MODULE 2: Knowledge Representation
Knowledge Representation Issues:
How to encode real-world knowledge into AI systems.
Challenges: Completeness, consistency, and reasoning efficiency.
Propositional Logic:
Basic logic with propositions connected by logical operators (AND, OR, NOT).
First-Order Predicate Calculus:
Extends propositional logic to handle objects, relations, and functions.
Uses quantifiers: Universal (For all) and Existential (There exists).
Horn Clauses:
A special type of logic formula used in logic programming and Prolog.
Resolution:
Rule-based inference method for deriving conclusions in logic.
Semantic Nets:
Graph-based representation showing relationships between objects.
Frames and Scripts:
Frames: Data structure representing stereotypical knowledge.
Scripts: Sequences of events used for common situations (e.g., restaurant visits).
Procedural vs Declarative Knowledge:
Procedural: 'How to' knowledge (e.g., steps to ride a bike).
Declarative: Facts and information (e.g., Paris is the capital of France).
Forward vs Backward Reasoning:
Forward Reasoning: Start from facts and apply rules to reach conclusions.
Backward Reasoning: Start from a goal and work backwards to find facts.
MODULE 3: Uncertain Knowledge and Reasoning
Quantifying Uncertainty:
Representing uncertain knowledge using probabilities.
Non-Monotonic Reasoning:
Logical reasoning where conclusions can be withdrawn if new knowledge is introduced.
Probabilistic Reasoning:
Using probability theories (Bayes' theorem) to deal with uncertainty in AI.
Certainty Factors:
Measures the degree of belief in a hypothesis.
Fuzzy Logic:
Deals with reasoning in situations where truth is not binary.
Truth values range between 0 and 1.
MODULE 4: Expert Systems
Need and Justification:
Expert systems provide decision-making capabilities for complex problems requiring human
expertise.
Knowledge Acquisition:
Process of gathering domain knowledge from experts into a system.
Architecture of Expert Systems:
Knowledge Base: Stores facts and rules.
Inference Engine: Applies rules to deduce conclusions.
User Interface: Communicates with the user.
Case Studies:
MYCIN: An expert system for medical diagnosis.
RI (Rete Inference): Framework for rule-based systems.
Learning:
Learning from Examples: Systems generalize rules from input data.
Knowledge in Learning: How prior knowledge helps improve learning.
Learning Probabilistic Models: Training models using probability distributions.
Reinforcement Learning: Learning through rewards and punishments (e.g., Q-learning).
