Machine Learning
... or how Google predicts
everything about you
By: Javier Fonseca
Back-end developer
What is Google?
Is it an Internet search company?
Is it an online
advertising service
company?
As per the Wikipedia definition, it is an Internet-related services company.
... a pretty broad definition, isn't it?
But what about their mission?
“Organize the world’s information and make it universally
accessible and useful.”
Source: https://www.google.com/intl/en/about/our-company/
How is Google able to achieve such a noble
goal?
● They store tons of information.
● They apply Machine Learning to it.
Some information Google tracks about users
● Searches
● "Ok Google" activity
● Web browsing activity
● Location history
● YouTube searches
● YouTube "Not Interested" feedback
● YouTube watches
● Mobile apps you used
● Google Ads settings
Source: https://myactivity.google.com/
How does Google use that information?
● To improve their services.
● And... to sell your personal data? NO!
● But... to make ads relevant.
Source: https://privacy.google.com/how-ads-work.html
Source: https://privacy.google.com/your-data.html
Okay okay...
... what is that machine
learning thing they do?
Machine Learning is...
"... the field of study that gives
computers the ability to learn without
being explicitly programmed"
Arthur Samuel, 1959
1959?!?!
Then why is it such a buzzword now?
Couple of reasons...
Computer
Speed
Source: Google (of course)
Emergence of Internet
Types of Machine Learning
● Supervised learning
● Unsupervised learning
Supervised Learning
Given labeled data, perform:
● Function Approximation
● Regression
● Classification
Supervised Learning
Common algorithms:
● Linear regression
● Logistic regression
● Support Vector Machines
● Naive Bayes
● Neural Networks
Unsupervised Learning
Given unlabeled data, detect:
● Patterns, structure
● Anomalies
● Clusters (groups)
Unsupervised Learning
Common algorithms:
● k-means
● Outlier detection
● Neural Networks
Neural Networks
Further Information
Software resources
● scikit-learn: machine learning in Python
● Anaconda: distribution of Python and R for large scale data
processing
● TensorFlow: An open-source software library for Machine
Intelligence
● Andrei Beliankou: Machine Learning with Ruby (curated list of links)
Further Information
Clouds
● Google Cloud Machine Learning
● Microsoft Azure Machine Learning
● Amazon AI
● IBM Watson
Further Information
Basic courses
● Adam Geitgey: Machine Learning is Fun! (8-part blog)
● Udacity Picodegree: a friendly introduction to Machine Learning
● Udacity: Intro to Machine Learning
● Google: Machine Learning recipes with Josh Gordon
● Coursera: Machine Learning
Further Information
Advanced Courses
● University of Stanford: CS231n Convolutional Neural Networks for
Visual Recognition
● fast.ai: Practical Deep Learning for Coders
● MIT Press: Deep Learning Book
● Udacity: Deep Learning by Google
● Coursera: Deep Learning specialization
Further Information
Communities, social networks
● Big Data Colombia (Slack)
● Machine Learning Colombia (Facebook Group)
● Kaggle Competitions
Further Information
And more...
Thank you!

Machine learning: Koombea TechTalks