Types of Machine Learning

Machine Leaning is the ability to make a model intelligent which in return can use the data and predict correct values efficiently.  

How to make the machines intelligent?

Here, the important factor to make any machine intelligent, by its own is to make the machine learn itself.

Machine learning algorithms fall into three categories:

Supervised Learning

• We have the training data in supervised learning.
• The name itself suggest supervised means a supervisor is present giving some instructions.
• That is the training data we have.

Training data 

Training data means we already have the input data as well as the output data, both available with us.

• On the basis of the input and the output data that is the training data or can also be called as labeled data, we create a model.
• In the model we give the inputs.
• Then, check that whether it's giving the valid output or not.

After giving the data to the Machine various learning algorithms comes in the role.

Example - 

Naïve Bayes Algorithm  

• Works on the basis of Supervised learning.
• Input and output data that is the training data, already made available.
• New inputs are given in the model during testing of the model.
• On the basis of the new inputs, given into the model, the model gives the output.
• If the output is correct that means the training data is good and a refined data.
• And, finally this means the algorithm properly learns and classifies on the basis of data. 

Unsupervised Learning

• Only input data is present no output data.
• Slowly and gradually the machine learns itself and goes towards the actual output.
• Uses Machine Learning algorithms to make decisions or outputs.
• Make groups or cluster for the unlabeled data.
• The algorithm further discovers patterns or groups in the data sets without human instructions.
• Has the ability to discover similarities and differences in the data.

Algorithms such as K- Mean clustering and market basket analysis follow unsupervised learning.

Reinforcement learning 

• Based on trial and error policy.
• Learns from the mistakes. 

Can be understand with an example - 

• Suppose an action is performed by the machine on the whole system.
• On the basis of the actions performed, certain plus points are awarded or a penalty if its error, provided theirs's a change in the state in the system.
• Now on the basis of these things the machine generates some policy.
• Next time the machine will act on the basis of the policy.