CHAPTER 3

Question 1

list the 4 ways to develop AI

Answer 1

1. heuristics
2. expert systems
3. machine learning
4. decision trees

Question 2

explain expert systems

Answer 2

an expert system is an algorithm which mimics the judgement of a and decision-making ability of a professional

4 steps
1. input
2. processes (changes made to the data)
3. structures (such as loops)
4. output

e.g.: someone is sick. the expert system can decide their illness from their symptoms.

Question 3

advantages of expert system

Answer 3

• there is no bias or favor
• because it is programmed, you can check if its judgement is fair
• you can consult an expert system if an expert is not around

Question 4

disadvantages of expert system

Answer 4

• not everything can be reduced to logical tests
• it is hard for an expert to put their skills into words

-expert systems are not always correct

Question 5

explain decision trees

Answer 5

a decision tree is a graph (like a dichotomous key) which uses a branching method to predict the output for given inputs

to automate a decision tree you can put it into a program and make use of if…else structure

decision trees can make human-like decisions

Question 6

explain machine learning

Answer 6

machine learning is a subfield of artificial intelligence characterized by its capability to mimic intelligent human behavior

Question 7

name the ways to train a computer

Answer 7

1. supervised training
2. unsupervised training
3. deep learning
4. reinforcement

Question 8

explain supervised training

Answer 8

in supervised training, the computer is given data that has already been organized and labelled (named)

Question 9

explain unsupervised training

Answer 9

in unsupervised training the computer is given a lot of data which has not been organized. The computer must find a pattern for itself.

Question 10

explain reinforcement training

Answer 10

reinforcement training tis the science of decision making. It learns from its mistakes.

Question 11

explain deep learning

Answer 11

deep learning can have powerful results as it combines the use of other methods of machine learning. Deep learning uses another kind of computer network called a neural network.

Question 12

advantages of machine learning

Answer 12

• you dont need to tell the computer how to solve the problem
• you can give the computer a problem and it’ll work out its own solution

Question 13

disadvantages of machine learning

Answer 13

-the computer needs a lot of data

• the data needs to be varied and diverse

-machine learning can go wrong

Question 14

list machine learning applications

Answer 14

machine learning is used in…

-diagnosing illness
-predicting the weather
-understanding speech
-spotting computer viruses

Question 15

explain heuristics

Answer 15

A Heuristics is a rule that helps you make a quick decision​

It is like a guess, or a rough estimate based on careful thinking about the problem​

It is not always completely accurate​

They provide a quick way to simplify difficult problems​

e.g.: you smell smoke, so you infer that theres a fire