3. Microsoft
Technology
Centers
The opportunity is
bigger than you
may think
$1.6Tdata dividend available
to businesses that embrace data
over the next four years
speed
More
people
New
analytics
Diverse
data
Data source: Microsoft & IDC, April 2014
14. Microsoft
Technology
Centers
What is Machine Learning ?
• Arthur Samuel (1959). Machine Learning: Field of study that gives computers
ability to learn without being explicitly programmed
• Tom Mitchell (1998). Well-posed Learning Problem: A computer program is said
to learn from experience E with respect to some task T and some performance
measure P, if its performance on T, as measured by P, improves with experience E.
• Chris Bishop (2006). The field of pattern recognition is concerned with automatic
discovery of regularities in data through the use of computer algorithms and with
the use of these regularities to take actions such as classifying the data to
different categories.
14
15. Microsoft
Technology
Centers
When is Machine Learning used?
• A pattern exists
• We cannot pin it down
mathematically
• We have data on it
• Learn it when you can’t code it
• Learn it when you can’t scale it
• Learn it when you have to
adapt/personalize
15
19. Microsoft
Technology
Centers
Ready to build a predictive model?
• How do I translate my business question to a
machine learning task? Or task?
• What data do I need to answer my question?
• Is my data clean?
• How much data do I need?
• Which model to choose? Are neural networks
better than random forests for my problem?
• Are my features right ones?
• Which features have the most predictive
value?
• My data has hundreds of attributes. What do
I do?
• How do I measure my model’s performance?
• When do I know I am done?
• My model performs perfect on my training
data but my client tells me its predictions are
terrible?
• How do I use my model from a smartphone
app?
• ………
19