Artificial Intelligence: Core Concepts and Problem Solving

Posted on Sep 27, 2026 in Computer Engineering

1. Understanding AI: Thought Process and Behavior

Artificial Intelligence (AI) can be understood based on thought process and behavior in four ways:

  • Thinking Humanly: AI tries to think like a human being. Example: A chatbot understanding a user’s question and giving a human-like response.
  • Thinking Rationally: AI uses logical reasoning to reach a conclusion. Example: An expert system using medical rules to identify a disease.
  • Acting Humanly: AI performs actions similar to humans. Example: A humanoid robot talking and interacting with people.
  • Acting Rationally: AI chooses the best possible action to achieve a goal. Example: A self-driving car selecting a safe route to reach a destination.

2. Cryptarithmetic Problem: BASE + BALL = GAMES

Solving the cryptarithmetic problem BASE + BALL = GAMES requires applying constraint satisfaction principles to assign unique digits to each letter.

3. Depth-Limited Search vs. Iterative Deepening DFS

Depth-Limited Search (DLS)

A variation of DFS where a fixed depth limit is specified. It does not explore nodes beyond that depth.

Iterative Deepening DFS (IDDFS)

Repeatedly performs depth-limited search by increasing the depth limit one level at a time.

Comparison

  • DLS: Has a fixed depth limit; may miss a solution beyond the limit; uses less memory.
  • IDDFS: Increases depth limit gradually; can find a shallow solution; no fixed limit required; complete for finite branching factors.

4. Converting Expressions to CNF

To convert (P → Q) → R into Conjunctive Normal Form (CNF):

  1. Remove implication: ¬(¬P ∨ Q) ∨ R
  2. Apply De Morgan’s law: (P ∧ ¬Q) ∨ R
  3. Apply distributive law: (P ∨ R) ∧ (¬Q ∨ R)

5. Genetic Algorithm Steps

Genetic Algorithm (GA) is an optimization technique inspired by natural selection:

  1. Initialization: Generate a random initial population.
  2. Fitness Evaluation: Calculate the fitness of each individual.
  3. Selection: Choose better individuals as parents.
  4. Crossover: Combine parents to produce offspring.
  5. Mutation: Apply random changes to maintain diversity.
  6. Replacement: Update the population.
  7. Termination: Repeat until a solution is found.

6. Learning: Inductive and Explanation-Based

Learning is the process of gaining knowledge from data to improve performance.

  • Inductive Learning: Obtaining a general rule from specific examples (e.g., observing that sparrows, crows, and pigeons fly leads to the rule “Birds can fly”).
  • Explanation-Based Learning (EBL): Learning a general rule by explaining why a specific example is correct using existing knowledge.

7. Machine Vision Steps

Machine vision enables computers to understand images through: Image Acquisition, Preprocessing, Segmentation, Feature Extraction, Pattern Recognition, and Decision Making.

8. AI Components

AI includes: Learning, Reasoning, Problem Solving, Knowledge Representation, NLP, Computer Vision, and Robotics.

9. Agent Types

An agent perceives its environment via sensors and acts via actuators. Types include: Simple Reflex, Model-Based Reflex, Goal-Based, Utility-Based, and Learning Agents.

10. Expert Systems

An Expert System uses expert knowledge to solve domain-specific problems. Development stages: Problem Identification, Knowledge Acquisition, Representation, Design, Implementation, Testing, Deployment, and Maintenance.

11. BFS vs. DFS

  • BFS: Explores level by level using a Queue. Finds the shortest path in unweighted graphs.
  • DFS: Explores as deep as possible using a Stack. Requires less memory.

12. Inference vs. Reasoning

Inference derives new facts from known rules, while Reasoning is the broader process of analyzing information to make decisions. Probabilistic reasoning is vital for handling uncertainty in real-world AI.

13. Goal-Based vs. Utility-Based Agents

A Goal-Based Agent focuses on achieving a specific state, whereas a Utility-Based Agent optimizes for the highest satisfaction (e.g., choosing a route based on time, cost, and safety).

14. Knowledge Representation Issues

Key issues include Representational Adequacy, Inferential Efficiency, and Handling Uncertainty. Solutions include using Bayesian Networks for uncertainty and logical validation for consistency.

15. Supervised vs. Unsupervised Learning

  • Supervised: Uses labeled data (e.g., spam detection).
  • Unsupervised: Uses unlabeled data to find patterns (e.g., customer clustering).

16. The Turing Test

Proposed by Alan Turing, this test evaluates if a machine can exhibit intelligent behavior indistinguishable from a human. Requirements include NLP, Knowledge Representation, Reasoning, and Learning.

17. Alpha-Beta Pruning

An optimization for the Minimax algorithm that prunes branches that cannot influence the final decision, using Alpha (MAX’s best) and Beta (MIN’s best) values.

18. Semantic Networks

A graphical representation where nodes are concepts and links are relationships. Useful for representing inheritance and hierarchical knowledge.

19. N-Queens Problem

The goal is to place N queens on an N×N board such that no two queens attack each other. It is typically solved using backtracking.

20. Artificial Neural Networks (ANN)

Inspired by the human brain, ANNs consist of Input, Hidden, and Output layers. Learning involves adjusting weights via Backpropagation to minimize error.

21. A* vs. Greedy Search

  • Greedy Search: Uses only h(n) (estimated cost). Fast but not optimal.
  • A* Search: Uses f(n) = g(n) + h(n). Guarantees an optimal solution with an admissible heuristic.

22. Natural Language Processing (NLP)

NLP enables computers to interpret human language. Pragmatic analysis is crucial for understanding the intended meaning behind literal statements based on context.