11. How did we
get here? 2016:
Google’s AI
beats
humans in
Go
1763:
foundations
of Bayes’
Theorem are
published
1950: Alan
Turing
proposes a
“learning
machine”
1956: the
term
“Artificial
Intelligence”
is coined by
John
McCarthy
1967: the
nearest
neighbor
algorithm is
created
1986:
backprop is
applied to
neural
networks
1995: first
work on
support
vector
machines is
published
28. Other Metrics
● Logarithmic Loss
● Area Under ROC Curve
● Mean Absolute Error
● … and many others!
Not sure what to track? Ask an expert!
29. When do I use
ML?
Complexity Scalability
Personaliz-ati
on
Adaptability
Available Secure Relevant Unbiased
Does your problem require…
Is your data…
30. How do I
launch an ML
product?
The same way you launch any other product!
Who is the
Customer?
What is the
problem?
What is the
benefit?
How do you know what Customers want?
Do you have data to support your idea?
What does the CX look like?
Consider:
31. How do I
launch an ML
product?
Is it possible to build an MLP without ML?
Build the MLP
and add
incremental
value with ML
Yes
No
Engage scientists
and tech teams
ASAP
32. Dos and
Don’ts
Do…
● Experiment!
● Think backwards from the
Customer
● Ask for advice from scientists
and ML experts
● Engage your tech teams early
and often
Don’t…
● Use ML without an
appropriate reason
● Forget to solve the problem
● Launch with “bad” data