Arrays (Copy)

Definition and Meaning of Artificial Intelligence (AI)

• Artificial Intelligence (AI):
  • A branch of computer science concerned with the design and development of systems that can perform tasks that normally require human intelligence.
  • Involves creating machines and software that can reason, learn, adapt, and make decisions.
  • AI systems can simulate intelligent behaviour and make independent decisions based on data and rules.
• Key Characteristics of AI:
  • Collection of data: The system gathers data from its environment or input sources.
  • Rules for using data: AI applies algorithms and logical rules to process data.
  • Ability to reason: AI can draw conclusions based on available data and rules.
  • Ability to learn: AI systems can improve performance over time using feedback and new data (machine learning).
  • Ability to adapt: AI can change strategies when faced with new or unexpected situations.
• Difference from traditional programming:
  • Traditional programs follow explicit step-by-step instructions.
  • AI systems can adapt their behaviour and sometimes create their own rules within defined parameters.

Main Branches Relevant to Syllabus

• Expert Systems (rule-based AI)
• Machine Learning (data-driven AI)

Expert Systems

Definition

• An expert system is an AI program designed to simulate the decision-making ability of a human expert in a specific domain.
• Uses a knowledge base and an inference engine to make recommendations or solve problems.

Main Components of an Expert System

1. Knowledge Base
   • Contains facts and information about the subject area.
   • Stores rules that relate facts together (e.g., IF–THEN rules).
   • Must be accurate, complete, and regularly updated.
   • Example: In a medical diagnosis system, knowledge base contains symptoms, diseases, and treatment rules.
2. Rule Base
   • Contains logical rules that describe relationships between facts in the knowledge base.
   • Example: IF temperature > 38°C AND cough = TRUE THEN possible illness = flu.
   • Rules are often written in if-then-else format.
3. Inference Engine
   • The reasoning mechanism of the expert system.
   • Applies rules to the knowledge base to deduce new facts or reach conclusions.
   • Can use forward chaining or backward chaining:
     • Forward chaining: Starts with known facts and applies rules to find new conclusions.
     • Backward chaining: Starts with a goal and works backward to determine if facts support it.
4. User Interface
   • The way users interact with the expert system.
   • Allows users to input queries or data and receive advice, diagnosis, or recommendations.
5. Explanation Facility (optional in syllabus but important in real systems)
   • Explains the reasoning behind decisions to users.
   • Increases trust in the system.

Operation of an Expert System – Step by Step

• Step 1: User enters problem details via the interface.
• Step 2: Inference engine retrieves relevant facts from the knowledge base.
• Step 3: Inference engine applies rules from the rule base to these facts.
• Step 4: New facts are generated or decisions are made.
• Step 5: The system outputs advice, solutions, or conclusions.
• Step 6: (Optional) The explanation facility shows how the decision was reached.

Examples of Expert Systems

• Medical diagnosis systems: Diagnosing illnesses based on symptoms.
• Weather forecasting systems: Predicting conditions based on atmospheric data.
• Mineral exploration systems: Identifying potential mining sites.
• Loan approval systems: Deciding if a loan should be granted.

Advantages of Expert Systems

• Can make consistent decisions without fatigue.
• Stores knowledge permanently and can be updated.
• Reduces reliance on availability of human experts.
• Can operate 24/7.
• Provides explanations for decisions.

Disadvantages of Expert Systems

• Knowledge base can become outdated without maintenance.
• Lacks human intuition and creativity.
• Cannot handle problems outside its programmed domain.
• Development can be expensive and time-consuming.

Machine Learning (ML)

Definition

• A subset of AI where systems learn from data without being explicitly programmed with specific rules for every possible situation.
• The system improves its performance as it is exposed to more data over time.

Basic Operation of Machine Learning

1. Data Collection
   • The system collects large amounts of relevant data.
   • Data must be accurate, relevant, and in the right format.
2. Data Preparation
   • Cleaning, organising, and formatting data for processing.
   • Removing errors, filling in missing values, normalising data.
3. Training
   • Feeding prepared data into a learning algorithm.
   • The algorithm finds patterns and builds a model representing relationships in the data.
4. Testing
   • Evaluating the model with new, unseen data to measure accuracy.
5. Deployment
   • Using the trained model to make predictions or decisions in real scenarios.
6. Feedback and Updating
   • Model is adjusted based on new data and performance results.

Types of Machine Learning

• Supervised Learning
  • Uses labelled data (input-output pairs).
  • The system learns to map inputs to outputs.
  • Example: Predicting housing prices based on features like size, location, and number of rooms.
• Unsupervised Learning
  • Uses unlabelled data.
  • The system tries to find hidden patterns or groupings.
  • Example: Customer segmentation based on purchasing behaviour.
• Reinforcement Learning
  • The system learns by trial and error, receiving rewards or penalties for actions.
  • Example: Game-playing AI that improves by winning points.

Examples of Machine Learning Applications

• Speech recognition: Translating spoken words into text.
• Image recognition: Identifying objects or faces in photos.
• Recommendation systems: Suggesting movies or products based on past behaviour.
• Autonomous vehicles: Learning to drive safely through simulations and real-world data.

Advantages of Machine Learning

• Can adapt to new data without reprogramming.
• Improves performance over time.
• Can process huge datasets quickly.
• Finds patterns humans might miss.

Disadvantages of Machine Learning

• Requires large amounts of high-quality data.
• Can produce biased results if training data is biased.
• Algorithms can be complex and not easily understood (“black box problem”).
• Training can require significant computing resources.

Comparison: Expert Systems vs Machine Learning

Applications of AI in Real Scenarios

• Industry:
  • Predictive maintenance of machinery.
  • Quality control through automated visual inspection.
• Transport:
  • Self-driving cars using AI to detect obstacles and make driving decisions.
  • Route optimisation in delivery services.
• Agriculture:
  • Crop disease detection using image recognition.
  • Precision farming to optimise fertiliser and water use.
• Medicine:
  • Automated analysis of medical images for diagnosis.
  • AI-assisted robotic surgery.
• Domestic:
  • Smart home assistants (e.g., Amazon Alexa, Google Assistant).
  • AI-powered cleaning robots.
• Entertainment:
  • Video game AI that adapts to player behaviour.
  • Streaming platforms recommending personalised content.