The three types of machine learning algorithms
VALENTIN RAKOVSKY, SABRINA BLANCHARD
Text size
Repeat
It corrects
its algorithm
to improve
its score
for its next
action
It assigns
a score
to its own
action
It assesses the
consequences
of its actions
based on
its state and
its environment
...then
performs
an action
The machine
makes a
decision...
5
4
3
2
1
The data in the same categories
can then be processed and
treated in a similar manner
The model receives data
without prior categorisation
Examples: video games, route
optimisation, training chatbots and
autonomous cars
Examples: customer segmentation,
recommendation systems,
image compression
It tries to group them
by similarity to
produce the most
homogeneous categories
Examples: anti-spam filters, disease
detection, image recognition
"Cat"
When provided with new data,
the AI recognises the patterns
of a known category
The model identifies
commonalities in the data
of the same category
Dog
The model is trained
on data categorised
by a human
Sources: Microsoft, IBM, Christina Bernard, Nvidia, Ai.nl
The three types of machine learning algorithms
The model is trained based on
its own performance and
and previous results
Reinforcement learning
Unsupervised learning