Machine Learning Basics - types of ML

Machine Learning is the set of techniques that helps to provide computers the capability to learn without being explicitly programmed. This is one of the most trending & an extra ordinary technologies that is changing computer world with a big & impactful extent. This from the name, it gives the computer the data value to programmed to make it more similar to humans: The ability to learn. Machine learning is actively being used today, perhaps in many more places than one would expect.

We can categories ML as in following 5 main types:

                         
Machine Learning  (Types)                     
                                                           
          
Supervised Learning      
Unsupervised Learning   
Reinforcement Learning       
Deep Learning        
Deep Reinforcement Learning
  
----- ML evolves from top to bottom ----

        

ML evolves from left to right — which means that computer scientists first started with supervised learning & then made enhancements in techniques & algorithms to achieve better came upto Unsupervised learning to Reinforcement learning then Deep Learning to Deep Reinforcement leaving. Still many enhancements are in progress with time …

Supervised Learning: Supervised learning is a type of machine learning with that an algorithm learns from labeled data to make predictions or classifications

Unsupervised Learning: With this type of ML technique we do not specify a target variable to the machine, rather we ask machine “What can you tell me about it?” We can provide a big data set to computer & can ask queries based on data. To solve or to give reply on these queries — the number of data points that the machine would require to deduce a strategy would be very large. 

For supervised learning, we can train the machine with even about few thousands of data points. But, in case of unsupervised learning, the number of data points that is reasonably accepted for learning starts in a few millions.

Reinforcement Learning: Reinforcement learning is a type of machine learning which enables AI-based systems to take actions in a dynamic environment through trial and error methods.

This was initially developed for machines to play games based on given an algorithm to analyze all possible moves (random) at each stage of the game. For a right move — the machine would be rewarded, otherwise it may be penalized. With the time machine starts differentiating between right and wrong moves and after several iterations would learn to solve the game puzzle with a better accuracy

Deep Learning: This ML type is a technique — model based on Artificial Neural Networks (ANN) — we can say convolutional Neural Networks (CNN)s. Deep learning uses several algorithms such as deep neural networks, deep belief networks, recurrent neural networks, and convolutional neural networks — these are been successfully applied in solving the problems of computer vision, speech recognition, natural language processing, bioinformatics etc. There are many fields where deep learning is proactively applied. 

This ML technique requires huge processing power and humongous data, which is in general easily available these days. 

Deep Reinforcement Learning: This ML technique — Deep Reinforcement Learning (DRL) is a combination of both deep and reinforcement learning. The reinforcement learning algorithms like Q learning are now combined with deep learning to create a powerful DRL model. Now a days this technique has been with a great success in the fields of robotics, video games, finance and healthcare.