AI for Marketing and Product Innovation

3 Predicting Using Big Data – Intuition Behind Neural Networks and Deep Learning

If there’s one thing the world’s most valuable companies agree on, it’s that their future success hinges on artificial intelligence.

– Enrique Dans, “Right Now, Artificial Intelligence Is the Only Thing That Matters,” Forbes, July 13, 2016

In this chapter, we look at the mathematical intuition behind algorithms of Machine Learning and AI is developed. The information is intended to be reader-friendly and present the core concepts for a non-mathematical audience that seeks to acquaint itself enough with the intuition to warrant further inquiry.

Intuition Behind Neural Networks and Deep Learning Algorithms

People are fond of saying that the brain is like a computer. But is it?

The answer: yes and no.

The fundamental rule about AI, one that has clobbered researchers over and over again since the 1950s, is this:

** The hard stuff is easy, but the easy stuff is hard. **

What does this mean? Well, for instance, computers can do arithmetic faster than any human could ever do, and they can remember an exponentially larger amount of information than a brain can retain, and they can do all of this at lightning speed, without a single error! That is, as long as they are programmed correctly.

If your program is incorrect, the computer can make the same error about a billion times in a minute, but at least it’s reliably unreliable, so the error is therefore findable and fixable.

Not only can computers ...