Technology journals are a buzzing over the revitalization of two technology concepts that have been stuck on the S-Curve of the Technology Life-Cycle for sometime now. The concepts: Machine Learning & Artificial Intelligence. So what do we really mean by Machine Learning & Artificial Intelligence(AI)? To the layperson it probably means machines than can actually “learn” (but they can’t, believe me they can’t). To my millennia’s it probably means something similar to what they saw in the movie iRobot. However, the truth is probably somewhere in between these two extremes. Let find out.
So what do we really mean by machine learning in 2017? Here is my definition: Machine Learning (ML) and Artificial Intelligence technologies as they exist today is a combination of computer science techniques, statistical methods and classification schemes harmonized to produce inference capabilities against (big) data sets. To be useful it has to be modeled into a system that has the ability to receive feedback from decisions and actions such that the model can be updated to produce better outcomes.
Approaches to machine learning can be categorized as: instance based learning, decision trees, perceptron’s and neural networks. All four (4) are worth further research. I will address these topics in subsequent posts.
There are many different applications for machine learning (ML) and artificial intelligence ranging from financial services and fraud detection to network security, medical diagnostics, predictive analytics and product recommendation systems. The application are end-less at this point. ML is no longer just for academia or large companies like Google, lBM, or Facebook. These algorithms are now mainstream and available for enterprise systems. It will be a disruptive force to reckon with if ignored. Cloud vendors will no-doubt start to offer AI and Machine Learning Services (actually they have already) to small, medium or large organizations.
This should be a no-brainer for organizations today.
Next question, what is the difference between supervised and unsupervised machine learning algorithms? Which is better? When should we use supervised? When should we used unsupervised? Stay tuned!