This document provides an introduction to machine learning. It begins by defining machine learning according to several experts as computational steps that can produce the best answer when applied iteratively to large data sets without being explicitly programmed. It distinguishes machine learning, deep learning, artificial intelligence, and discusses their relationships with data science and big data. The document outlines different types of machine learning including supervised, unsupervised, regression, and classification. It also discusses implementing machine learning solutions, neural networks, and provides suggestions for further research.

1. MACHINE
LEARNING
AN INTRODUCTION
Software Leadership of Kalamazoo
Aug 2018

2. SLOK ADMINISTRATION
• Peer to Peer Mentoring presentation moved to September (Brian White)
• Standardize Meetings on the last Tuesday of Every Month?
• SLoK Facebook Page
• Meetups name change to SLoK Talks
• Marketing
• Goal is to have 30 average attendees
• Need ~ 140 members
• We’re all marketers for this group
• Need New Presenters

3. FEEDBACK
• What do you think Machine Learning is?
• Terminator or Ex Machina?
• Clothes folding robot?
• Artificial Intelligence ≠ Robotics
• Robotics commonly employ AI logic, but the two are not the same
• Despite what Hollywood is trying to tell us

4. DEFINITIONS
Arthur Samuel: The field of study that gives computers the ability to learn without being explicitly
programmed.
Tom Mitchell: A computer program is said to learn from experience E with respect to some class of
tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves
with experience E.
Kevin Jackson: Computational steps, that when applied iteratively to a large set of data, will
produce a “best” answer on its own.

5. Deep Learning
Imitate the human brain with multilayer neural
networks.
Examples: Natural speech recognition (Siri,
Cortana), hand writing recognition, newer
automated driving systems
Machine Learning
Systems that “learn” (get
better on their own) given
a large data set (training).
Examples: Recommender
systems (Netflix, Amazon),
Prediction systems (House
values), Anomaly Detection
(Credit card fraud)
Artificial Intelligence
Any technique that
allows computers to
mimic human intel-
ligence. Does not have
to “learn” over time.
Examples: Most
programs that make a
decision based the input
they have. May even
include complex if-
then statements
Artificial Intelligence(AI) vs Machine Learning(ML) vs Deep Learning(DL)

6. Machine
Learning
Big
Data
Data
Science
Relationship between Machine Learning, Data Science, and Big Data

7. TYPES OF MACHINE LEARNING
• Supervised – We are given a set of features and expected results
• Regression – Predict results within a continuous output. Map input variables to some
continuous function
• Classification – Map input variables to discrete categories
• Unsupervised – Little or no idea of what the expected results should look like
• Clustering – Automatically group data into clusters or categories
• Non-Clustering – Messy data, such as audio cleaning (the cocktail party algorithm)

8. Types of Machine Learning with Common Models

9. IMPLEMENTING MACHINE LEARNING SOLUTIONS
• Gradient Descent, Mean Squared Errors, and the Cost Function
• Review
• For each feature x, with an expected result y, we “train” a variable Ɵ
• MatLab (example)
• The math gets involved quickly (linear algebra, calculus derivatives)
• Luckily, platforms are being created to make this much less complex
• Tensorflow, PyTorch, etc
• Azure ML Studio (example)
• Programming language libraries
• Python, R, others catching up

10. NEURAL NETWORKS
• Neural networks are a set of algorithms, modeled loosely after the human brain, that are designed to recognize patterns. They
interpret sensory data through a kind of machine perception, labeling or clustering raw input. The patterns they recognize are
numerical, contained in vectors, into which all real-world data, be it images, sound, text or time series, must be translated
• Example

11. FURTHER RESEARCH
• Types of Machine Learning
• Discussion of Gradient Descent and basic math used in regression analysis
• Examples of how Machine Learning is being used in industry
• Excellent Machine Learning course through Stanford on Coursera
• Microsoft Azure Machine Learning Studio (free subscription)
• Neural Network overview