Chapter 3: Solving Problems by Searching Flashcards
(15 cards)
What is an agent that needs to plan ahead called?
A problem-solving agent
Problem-solving agents consider sequences of actions to achieve a goal state.
What is the computational process that problem-solving agents undertake?
Search
Search is the process of finding a path to a goal state.
What type of representations do problem-solving agents use?
Atomic representations
Atomic representations consider states of the world as wholes without internal structure.
What are agents that use factored or structured representations of states called?
Planning agents
Planning agents are discussed in Chapters 7 and 11.
What types of environments does this chapter consider for search algorithms?
Episodic, single agent, fully observable, deterministic, static, discrete, and known
These constraints simplify the search algorithms being discussed.
What distinguishes informed algorithms from uninformed algorithms?
Informed algorithms can estimate how far they are from the goal; uninformed algorithms cannot
Informed algorithms have access to additional information about the goal.
What notation is used to describe asymptotic complexity in this chapter?
O(n) notation
Readers unfamiliar with asymptotic complexity should consult Appendix A.
What is the first phase of the problem-solving process for an agent?
Goal formulation
Goal formulation organizes behavior by limiting objectives and actions.
What does the agent do in the problem formulation phase?
Devises a description of the states and actions necessary to reach the goal
This includes creating an abstract model of the relevant part of the world.
What is the purpose of the search phase in the problem-solving process?
To simulate sequences of actions and find a sequence that reaches the goal
The sequence found is called a solution.
What is a solution in the context of a problem-solving agent?
A sequence of actions that reaches the goal
For example, the sequence from Arad to Bucharest.
What is an important property of solutions in a fully observable, deterministic, known environment?
The solution is a fixed sequence of actions
The agent can ignore percepts while executing the actions if the model is correct.
What is an open-loop system in control theory?
A system that ignores percepts during execution of actions
This approach is safe in fully observable environments where the model is accurate.
In partially observable or nondeterministic environments, what must a solution include?
A branching strategy
The branching strategy recommends different future actions based on incoming percepts.
Fill in the blank: The agent might plan to drive from Arad to Sibiu but may need a _______ in case it arrives in Zerind by accident.
contingency plan
This illustrates the need for adaptable strategies in uncertain environments.
