Artificial Intelligence - Detailed Notes
Module 5: Planning and Learning
5.1 The Planning Problem:
Planning in AI is the process of selecting and organizing actions to achieve a desired goal. It involves
deciding "what to do" before doing it. Planning operates over a state space defined by an initial state, goal
states, and available actions with defined preconditions and effects.
Types of Planning:
- State Space Search: Finding paths through a space of states.
- Forward Search: Progresses from initial to goal state.
- Backward Search: Works backwards from goal to initial state.
- Partial Order Planning: Allows partial ordering of actions using the least commitment strategy.
- Hierarchical Planning: Breaks complex tasks into simpler sub-tasks using a task hierarchy.
- Conditional Planning: Plans with branches depending on possible future conditions in uncertain
environments.
5.2 Learning:
Learning in AI enables agents to improve from experiences and data.
Forms of Learning:
- Supervised Learning: Learns from labeled examples to predict outcomes.
- PAC Learning: A theoretical model analyzing the effectiveness of learning algorithms.
- Reinforcement Learning: Learns via interaction with the environment using reward-based feedback.
Types of Reinforcement Learning:
- Passive Reinforcement Learning: Evaluates a fixed policy.
- Active Reinforcement Learning: Improves the policy through exploration and feedback.
Artificial Intelligence - Detailed Notes
Module 6: AI Applications
A. Natural Language Processing (NLP):
NLP is the ability of computers to understand and process human language. Key components include:
- Language Models: Predict the next word using statistical or neural models.
- Grammars: Define valid sentence structures using production rules (e.g., context-free grammars).
- Parsing: Analyzes sentence structure. Methods include top-down and bottom-up parsing.
B. Robotics:
Robotics integrates AI with mechanical systems. A robot consists of sensors, actuators, and a controller.
- Sensors: Perceive the environment (e.g., cameras, LIDAR).
- Actuators: Perform physical actions (e.g., motors).
- Controller: Makes decisions.
Common tasks: navigation, object recognition, interaction, automation.
C. AI Applications in Domains:
- Healthcare: Diagnosis, treatment recommendations, medical image analysis.
- Retail: Recommendations, inventory management, chatbots.
- Banking: Fraud detection, loan automation, AI advisors.
