Please enable JavaScript.
Coggle requires JavaScript to display documents.
Methods computers use to solve problems - Coggle Diagram
Methods computers use to solve problems
Heuristics
Quick decisions
Problem solving
Guess / Rough estimate
Humans use heuristics, so this helps the machine think like a human
Advantages
Quickly solution of problems
Less work for the user
Good results
Disadvantages
Doesn't always give the best solution
Not always accurate
Used by virus-checking softwares
Decision trees
Shows all the decisions
Each branch stands for a different choice
By following the branches the right decision is got
In square boxes
Automated
Works on its own without a human user
Works with if...else structures
Expert system
Based on knowledge
Human-like decisions
By taking all the factors a human would consider
Four components
Processes
Changes made to the input
Structures
Loop and if
Outputs
Final product
Inputs
Data used
It can:
Determine jail sentences
Make a diagnosis of
A broken car
An ill patient
Advantages
The rules will always be fair
It can decide without prejudice or favour
When an expert is not available it can be used
Disadvantages
Not every decision can be made by logical tests
Experts can be wrong, and if they are, the algorithm is too
It can be hard to put an expert's skills into words
Machine learning
It works different from an algorithm
Training
The computer learns for itself how to solve problems
Types:
Supervised training
The data that's given to the computer is already organised and labelled
Unsupervised training
The data is not sorted nor organised
The computer has to find the patterns fot itself
Reinforcement training
The computer produces random or undirected signals or actions
Reinforcement is feedback that lets the computer know if it has moved towards the right goal
The computer learns to produce output that's closer to the right result
Deep learning
Combines the other methods into a highly complex learning process
Neural network
It is a different type of computer structure
Produces very powerful results
Advantages
It is not needed to tell the computer how to solve a problem
The computer is given a goal and it will work out its own solution
Disadvantages
The computer needs a lot of data
The data must be diverse and varied
It can go wrong
Other purposes
Diagnosisng illnesses
Weather forecasting
Understanding speech
Spotting computer viruses