Artificial Intelligence & Soft Computing Notes
Subject Overview
The first Elective-II option, covering classical AI — problem formulation, knowledge representation via propositional and first-order logic, heuristic search strategies and natural language processing — and soft computing — neural networks, fuzzy systems and genetic algorithms. A 4-credit elective theory paper.
Subject at a glance
4
Units
39
Topics
4
Credits
4
Infographics
Unit-wise Syllabus
4 units — open any unit for detailed topic-wise notes, diagrams and important exam questions
- Foundations and history of AI
- toy and real-world problems (Tic-Tac-Toe, Water Jug, 8-puzzle, 8-Queens)
- problem formulation and search
- propositional logic and theorem proving
- resolution
- Horn clauses
- forward/backward chaining
- first-order logic and inference
- 1Convert sentences to clause form (CNF)
- 2Negate the goal and add it
- 3Resolve pairs of clauses
- 4Add the resolvents
- 5Empty clause found
Goal is proved
Study guide
This unit covers the foundations and history of AI, classic problems (tic-tac-toe, water jug, 8-puzzle, 8-queens), search formulation, propositional logic and theorem proving, resolution, Horn clauses, chaining and first-order logic. Practise resolution proofs step by step.
- Hill climbing
- simulated annealing
- greedy best-first search
- A* and optimal search
- memory-bounded heuristic search
- NLP grammars
- parsing
- semantic analysis and pragmatics
Hill climbing
Yes
No: can get stuck at local maxima
Simulated annealing
Yes
Can escape local maxima
Greedy best-first
Yes: h(n)
No
A*
Yes: f(n) = g(n) + h(n)
Yes, if h is admissible
Study guide
You study heuristic search: hill climbing, simulated annealing, greedy best-first, A*, memory-bounded search, and natural language processing: grammars, parsing, semantic analysis and pragmatics. Trace A* on a small graph.
- Soft computing vs hard computing
- major areas and applications of soft computing
- neural network learning rules and activation functions
- single-layer perceptrons
- backpropagation networks and architecture
- associative memory
- adaptive resonance theory
- self-organizing maps
- unsupervised learning networks
- 1Forward pass
Compute outputs layer by layer
- 2Compute error
Target − output
- 3Backward pass
Propagate error gradients
- 4Update weights
Δw = η × δ × input
- 5Repeat for each epoch
Study guide
You study soft computing versus hard computing and neural networks: learning rules, activation functions, perceptrons, backpropagation, associative memory, adaptive resonance theory and self-organising maps. Work through one backpropagation iteration by hand.
- Fuzzy set theory
- fuzzy vs crisp sets
- fuzzy relations
- fuzzification
- min-max composition
- defuzzification
- fuzzy logic and rule-based systems
- fuzzy decision making and control
- genetic algorithm history
- encoding methods
- fitness functions
- GA operators (reproduction, crossover, mutation)
- convergence
- introduction to hybrid systems
- 1. Initial population: Random encoded solutions
- 2. Fitness evaluation:
- 3. Selection: Fitter survive
- 4. Crossover: Combine parents
- 5. Mutation: Random changes
Study guide
You study fuzzy sets, relations, fuzzification, min-max composition, defuzzification and fuzzy control and genetic algorithms: encoding, fitness, reproduction, crossover, mutation, convergence and hybrid systems. Design a simple fuzzy controller.
Why Artificial Intelligence & Soft Computing Matters
AI and soft computing underpin modern intelligent systems from recommendation engines to control systems, and form the base for ML, deep learning and AI careers.
How to Prepare for Artificial Intelligence & Soft Computing
- Trace search algorithms and neural network updates by hand.
- Implement a small genetic algorithm.
- Compare fuzzy logic with crisp logic using examples.
Artificial Intelligence & Soft Computing: Frequently Asked Questions
What is soft computing?
Soft computing is a set of techniques, including neural networks, fuzzy logic and genetic algorithms, that handle imprecision and uncertainty to solve complex problems.