Aii Test

Question 1

What is the Hillclimbing problem and how does it work?

Answer 1

It’s the simplest AI algorithm.

It guesses random variables, runs the algorithm, assesses the outcome and repeats the process with the lates/ best outcome

Question 2

What is the negative with the Hillclimbing approach and how can you over come it?

Answer 2

Making random decisions therefore could get stuck in a local minimum

You can over come it with simulated annealing

Question 3

What is reinforcement learning?

Answer 3

-Reinforcement learning techniques establish a mapping of situations to actions so as to maximize a
numerical reward signal.
-Explicitly considers the whole problem of a goaldirected agent interacting with an unknown/uncertain
environment.
-Learning through trial and error.

Question 4

What do the terms state, action, reward function, value function and policy refer to in reinforcement learning?

Answer 4

State- current state an environment is in and sends it to the agent
Action- Agent will act upon the input from state in the environment
Reward- feedback is given after the action is taken
Value-The value function estimates the long-term desirability or utility of being in a particular state. It represents the expected cumulative reward an agent can achieve from a given state by following a specific policy.

Policy- A policy is a strategy or rule that determines the agent’s behavior in a given state. It maps states to actions and guides the agent’s decision-making process.

Question 5

What are Swarm Algorithms?

Answer 5

-A computational Method inspired by the behaviour of many living creatures
-They solve complex algorithms by simulating the collective behaviour of a group of simple agents (particles) that interact with each other and their environment

Question 6

What is Supervised Learning?

Answer 6

-Learning from examples provided by a knowledgeable
external environment (supervisor).
- Environment directly indicates what the action should
have been.
-Such instructive feedback is independent of the output
actually given by the agent – no need to explore the
environment (passive).

Data- build model from data- get output- output adjust model

Question 7

What is Unsupervised Learning?

Answer 7

 Unsupervised learning aims to capture the
structure of unlabelled data.
 The most well-known form is termed clustering.
 A cluster is a collection of data items that are
similar to each other and dissimilar to data
items in other clusters.