In my position at IBM, I regularly brief executives, managers, and computer
professionals on data mining and neural network technology. In my briefings,
I cover the fundamentals of data mining and neural networks, and I
also discuss specific applications relevant to the customers' businesses.
Since time is usually limited, my goal is to quickly give them a basic understanding
of data mining and to spark their imaginations so they can visualize
how the technology can be used in their own enterprises. When I succeed, it
is satisfying to see their excitement as they "ponder the possibilities." In the
question-and-answer period following my presentations, I am invariably
asked for a recommendation on a "good book on neural networks" so they
can learn more. With few exceptions, these people do not want to know how
neural networks work; they want to know how neural networks can be applied
to solve business problems, using terminology they can understand
and real-world examples to which they can relate.
While there are many neural network books available today, most focus on
the inner workings of the technology. These texts approach neural networks
from either a cognitive science or an engineering perspective, with a corresponding
emphasis either on philosophical arguments or on a detailed treatment
of the complex mathematics underlying the various neural network
models. Other neural network books discuss academic applications, which
have little or no relation to real business problems, and are full of C or C++
source code showing nitty-gritty implementation details. None of these titles
would fit my definition of a "good book on neural networks" that is appropriate
for a business-oriented audience.