AI for Marketing and Product Innovation

6 Core Algorithms of Artificial Intelligence and Machine Learning Relevant for Marketing

A real artificial intelligence would be intelligent enough not to reveal that it was genuinely intelligent.

– George Dyson (attributed), “Enthusiasts and Skeptics Debate Artificial Intelligence,” Vanity Fair, November 26, 2014

Machine Learning algorithms come in a few flavors, and typically solve problems by learning from data, without human intervention.

Artificial Intelligence codifies rules used by experts, and mimics human decision-making.

In this chapter, intuition regarding Machine Learning algorithms is provided. The goal is not so much to describe the formulae but rather the core idea behind the method.

While algorithms can be classified by many different schema, the most common ways of classifying them are the following. These schema are based on the level of active human intervention required in fine-tuning the algorithms:

• Supervised Learning
• Unsupervised Learning
• Reinforcement Learning

Supervised Learning

Basically, Supervised Learning refers to algorithms that create a black box model between inputs and output, but are “supervised” in the following manner – they are told in training what the “right answers are” and what the “wrong answers are.” So they build a model between inputs and outputs always trying to prune the model so the inputs could rightly predict the right answers and avoid the wrong ones.

Y = f(X)

where f equals function. If the mapping function guesses well ...