Machine Learning Vs AI Vs Advanced Analytics Vs Deep Learning

There is a lot of buzz around Machine Learning (ML) and Artificial Intelligence (AI) across all industries and for good reasons. Over the last 18 months we have spent time exploring and understanding the domain for its implementations and benefits at TUI.  With the creation of the thomson chatbot and the recommendation engine for search we do have some experience in field now but surely there is a lot more to do. This blog (and hopefully a few more to come) is an attempt to share some of the knowledge and experience gathered. 

So how is this different from Advance Analytics, AI or Deep Learning  ?

We often come across these terms Machine Learning, Advanced Analytics, Artificial Intelligence, Deep Learning and have been using them interchangeably as well. However we can point at least share some difference and use the right term at the right place.  

Advanced Analytics normally produces an output in the form of a file which needs to be consumed manually.  

Machine Learning systems  -  take it a step further and converts it into a system by automating the input, processing and output. The systems normally produces the same type of output eg. product recommendation system will always produce product recommendation. if we need the prediction of does a patient have a cancer or not then the minimum we have to do is to re-engineer the input and validate the output. The system cannot do it on its own. ie. the Learning and output are specific to data and need human intelligence to engineer it.  Each use case also has a different algorithm that give max efficiency for that particular problem. 

These systems are most common and have a wide variety of usage i.e spam filters, churn analysis, recommendation systems, chatbots, predictive maintenance. gaming strategies. high frequency trading on stock markets. quantitative portfolio management in investment banking etc.  

Deep Learning is a special branch of Machine Learning which uses the Neural networks as opposed to Regression, Classification or Clustering. This is the technique used for image recognition or cancer predictions. So when facebook identifies a face in a picture its using deep learning behind it. 

Artificial Intelligence or General Intelligence - AI is the next step for Machine Learning. A domestic robot has to do multiple jobs i.e dishes, cooking, cleaning, laundry, security, web search, shopping list etc. Similarly an Artificially intelligent office assistant, would have answers to all Management information, can optimize pricing based on users preference, handle churn, yield, orders, finance and every other operation single handedly. If we choose the path to coding specific tasks using specific algorithms and bundle them together, it will soon hit the complexity problem.  General Intelligence or AI is the goal of producing a master algorithm capable of learning everything using a single algorithm. 

It would still not have a will of its own! No HAL :)