Postgrad Notes / PGD Computer Science / First Semester / CMS 707 Artificial Intelligence / Knowledge Map

Download .md

Knowledge Map: Artificial Intelligence (CMS707)

Generated: 2026-04-02 | Topics: 13 | Concepts: 34 | Relationships: 31

Course Overview

CMS707 covers the theory and practice of Artificial Intelligence, from the foundational question of what intelligence is, through problem-solving via state space search, to knowledge representation using predicate logic and probabilistic reasoning. The course then extends to advanced application areas including machine learning, NLP, expert systems, fuzzy logic, and genetic algorithms. It is grounded in the AIMA (Russell & Norvig) framework and closely follows the lecturer's lecture series.

Concept Hierarchy

Artificial Intelligence (CMS707)
│
├── PART I: Problem-Solving and Search
│   │
│   ├── 1. Introduction to AI                          [5 concepts]
│   │   ├── Key Concepts: Intelligence, Artificial Intelligence, Mundane Tasks, Formal Tasks, Turing Test
│   │   ├── Central: Artificial Intelligence
│   │   └── Note: Mundane tasks (perception, speech) are harder to compute than formal tasks (chess, logic)
│   │
│   ├── 2. AI Agents and Rational Agents               [6 concepts]
│   │   ├── Key Concepts: Agent, Rational Agent, Percept Sequence, Agent Function, Agent Program, Performance Measure
│   │   ├── Central: Rational Agent
│   │   └── Note: Agent Function (abstract math) → implemented by → Agent Program (concrete code)
│   │
│   ├── 3. Problem-Solving and Search                  [10 concepts]
│   │   ├── Key Concepts: State Space Representation, BFS, DFS, Iterative Deepening Search,
│   │   │                 Best-First Search, A* Search, Heuristic, Water Jug Problem, Farmer Problem
│   │   ├── Central: State Space Search, BFS/DFS
│   │   └── 4-step method: Define → Analyse → Isolate/Represent → Choose technique
│   │
│   └── 4. Heuristic Search Techniques                 [3 concepts]
│       ├── Key Concepts: Heuristic, Informed Search, Uninformed Search
│       └── Best-First = hybrid of BFS completeness + DFS memory efficiency
│
├── PART II: Knowledge Representation and Reasoning
│   │
│   ├── 5. Knowledge Representation                    [5 concepts]
│   │   ├── Key Concepts: Natural Language KR, Simple Relational KR, Inferential KR,
│   │   │                 Hallucination, Facts vs Representations
│   │   └── Central: Inferential KR (supports inheritance + logical inference)
│   │
│   ├── 6. Predicate Logic                             [6 concepts]
│   │   ├── Key Concepts: Predicate Logic, Quantifiers (∀/∃), Constants/Variables/Predicates,
│   │   │                 Inference Rules, Resolution, Forward/Backward Chaining
│   │   ├── Central: Predicate Logic
│   │   └── Example: Marcus/Caesar, 8 axioms → logical inference chain
│   │
│   ├── 7. Probabilistic Reasoning                     [3 concepts]
│   │   ├── Key Concepts: Uncertainty, Bayesian Reasoning, Bayesian Networks
│   │   └── Requires: Predicate Logic as prerequisite
│   │
│   └── 8. Semantic Nets                               [3 concepts]
│       └── Key Concepts: Semantic Networks, Frames, Scripts
│
└── PART III: Advanced Topics and Application Areas
    │
    ├── 9. Learning                                     [6 concepts]
    │   └── Supervised Learning, Unsupervised Learning, Reinforcement Learning,
    │       Decision Trees, Neural Networks, Deep Learning
    │
    ├── 10. Natural Language Processing                 [4 concepts]
    │   └── Speech Recognition, Machine Translation, Question Answering (Watson), Language Models
    │
    ├── 11. Expert Systems                              [2 concepts]
    │   └── Knowledge-Based Systems, Inference Engines
    │
    ├── 12. Fuzzy Logic Systems                         [2 concepts]
    │   └── Fuzzy Sets, Fuzzy Inference
    │
    └── 13. Genetic Algorithms                          [2 concepts]
        └── Selection, Crossover, Mutation

Key Relationships

Hierarchical (Is-a / Part-of)

Causal / Enables

Dependency / Prerequisite

Contrast (Contrasts-with)

Combines / Synthesises

Solves

Central Concepts

These are the most connected concepts in the course, master these first:

Concept Connections Why Central
Predicate Logic 4 Enables KR, used in inference, expresses inferential KR, bridges natural language and reasoning
Rational Agent 4 Defined by goal maximisation, evaluated by performance measure, acts on percept sequence, foundation of all AI system design
State Space Search 4 Foundation of problem-solving, solves classic problems, prerequisite for BFS/DFS/A*, underpins all search algorithms
BFS / DFS 5 Combined by Iterative Deepening and Best-First, contrasted by memory/completeness trade-offs, basis of all uninformed search
Machine Learning 3 Basis of Deep Learning, Reinforcement Learning, and Supervised Learning, connects to NLP, Robotics, Expert Systems
Knowledge Representation 5 Subsumes Natural Language, Relational, Inferential KR; enables Expert Systems; tied to NLP and Predicate Logic

Prerequisite Learning Path

Recommended study order based on concept dependencies:

  1. Intelligence & AI: What is AI? What counts as intelligent behaviour?
  2. Mundane vs Formal Tasks: Why mundane tasks are computationally harder
  3. Agents & Rational Agents: Percept Sequence, Agent Function, Performance Measure
  4. State Space Representation: Foundation of all problem-solving
  5. Uninformed Search (BFS, DFS): How to navigate state spaces
  6. Informed Search (Heuristic, Best-First, A)*, Smarter navigation
  7. Iterative Deepening: Combining BFS completeness + DFS memory
  8. Knowledge Representation (Natural Language → Relational → Inferential): Moving from data to reasoning
  9. Predicate Logic: Formal language for inference; Marcus/Caesar worked example
  10. Probabilistic Reasoning & Bayesian Networks: Handling uncertainty
  11. Semantic Nets: Graph-based KR
  12. Machine Learning (Supervised → Unsupervised → Reinforcement → Neural Networks → Deep Learning)
  13. Application Areas (NLP, Expert Systems, Fuzzy Logic, Genetic Algorithms, Robotics)

Definitions Quick Reference

Term Definition (brief)
Intelligence Capacity to think and comprehend the world, involving knowledge, experience and understanding
Artificial Intelligence Field of building systems that act rationally to maximally achieve predefined goals
Agent Entity that perceives its environment via sensors and acts via actuators; persists and adapts
Rational Agent Agent that acts to achieve the best/expected outcome, judged by a performance measure
Percept Sequence Complete history of everything an agent has ever perceived
Agent Function Abstract mapping from percept sequence to action
Agent Program Concrete implementation of the agent function
Performance Measure Criterion capturing best/expected outcome for a rational agent
State Space Set of all possible states in a problem, connected by operators
BFS Breadth-First Search, FIFO queue, explores level by level, complete and optimal
DFS Depth-First Search, LIFO stack, memory efficient, may get trapped
Iterative Deepening DFS with increasing depth limit; BFS completeness + DFS memory
Best-First Search Heuristic-guided search combining BFS and DFS advantages
A* Search Informed search using f(n) = g(n) + h(n); optimal if heuristic admissible
Heuristic Problem-specific knowledge estimating cost/distance to goal
Knowledge Representation Mapping of real-world facts into a formalism that can be manipulated
Predicate Logic Formal system with quantifiers (∀, ∃), predicates, constants, variables
Hallucination (AI) Representation of facts that may not exist; absence of self-aware grounding
Bayesian Reasoning Probabilistic inference using prior knowledge and conditional probabilities
Fuzzy Logic Truth values between 0 and 1; handles vagueness and degrees of truth
Genetic Algorithm Evolutionary optimisation via selection, crossover, and mutation
Turing Test Benchmark: can a machine converse indistinguishably from a human?

Tutor Reference Notes

For use by the tutor agent during learning sessions with Mbosinwa: