@fellyph
Using Machine Learning to
improve the user experience
Fellyph Cintra
Slide 2
Slide 2 text
@fellyph
Or: Machine learning for
mortal developers
Slide 3
Slide 3 text
@fellyph
@fellyph
Deloitte Digital
Google Developer Expert
Slide 4
Slide 4 text
@fellyph
Artificial Intelligence ?
Slide 5
Slide 5 text
@fellyph
Artificial Intelligence ???
Slide 6
Slide 6 text
@fellyph
–Arthur Samuel
“Field of study that gives computers the ability to
learn without being explicitly programmed.”
Slide 7
Slide 7 text
@fellyph
250
300
100
270
4
151
Example: Normal computation
If (x > 200) {
side = right;
} esle {
side = left;
}
Right - 250
Right - 300
Left - 100
Right - 270
Left - 4
Left - 151
Slide 8
Slide 8 text
@fellyph
Example: AI computation
Right - 250
Right - 300
Left - 100
Right - 270
Left - 4
Left - 151
Predict next
results
Slide 9
Slide 9 text
@fellyph
Example: AI computation
Right - 250
Right - 300
Left - 100
Right - 270
Left - 4
Left - 151
Predict next
results
Slide 10
Slide 10 text
@fellyph
AI Terms
Natural Language
Neural network
Datasets
Supervised Learning
Unsupervised Learning
Pre-trained models
Machine Learning
Deep Learning
Slide 11
Slide 11 text
@fellyph
AI Terms
Natural Language
Neural network
Datasets
Supervised Learning
Unsupervised Learning
Pre-trained models
Machine Learning
Deep Learning
Slide 12
Slide 12 text
@fellyph
Artificial Intelligence
Machine Learning
Deep Learning
Natural
Language
Neural
Network
Supervised
Learning
Unsupervised
Learning
Slide 13
Slide 13 text
@fellyph
Slide 14
Slide 14 text
No content
Slide 15
Slide 15 text
@fellyph
ML5.js
Slide 16
Slide 16 text
@fellyph
What we can Classify?
Images Sounds Text
Slide 17
Slide 17 text
@fellyph
Can we use ML5.js and
WordPress?
Slide 18
Slide 18 text
@fellyph
Ναί!!!
Slide 19
Slide 19 text
@fellyph
Steps
1. Create a Gutenberg Block(optional)
2. Define labels(output) - Product ID
3. Training model
4. Export Pre-trained model
5. Apply script to a Gutenberg block
Slide 20
Slide 20 text
@fellyph
LET’S CODE!
Slide 21
Slide 21 text
@fellyph
MAIN FUNCTIONS
Slide 22
Slide 22 text
@fellyph
Training
Slide 23
Slide 23 text
@fellyph
Solution
Slide 24
Slide 24 text
@fellyph
gotResults ruction
Slide 25
Slide 25 text
@fellyph
Solution
Slide 26
Slide 26 text
@fellyph
Considerations
• KNN model - 500kb
• FeatureExtractor model - 5MB(more than 2 labels).