Page 56 - Reclaim YOUR DIGITAL GOLD (without audio)

RECLAIM YOUR DIGITAL GOLD

If we want machines to behave and think like humans,
we must first understand how humans learn. This will
enable us to create machines that act and think like
humans.

The obvious answer to the question of how humans
learn is that we learn by processing data. Infants learn to
walk and talk by absorbing a large amount of information
(data) and processing it to recognize similarities and
patterns, which may apply.

With that in mind, let’s go over a simple example of how
data collection works in machine learning, and then we’ll
use that as an opportunity to discuss the steps involved
in using machine learning to draw conclusions from the
data.

Assume we’ve been tasked with developing a method for
determining whether a beverage is beer or wine. We would
start with the question: “What is the difference between
wine and beer?” Then we would attempt to answer that
question using data. This question-answering system
that we are developing is known as a “model,” and the
process by which we generate this model is known as
“training.” The goal of training is to create a dependable
model that can respond to our inquiries correctly the
majority of the time. However, before we can train a
model, we must first collect data to use as training
material. This is the point where we begin.

36