Question

What is meant by the term overfitting in the context of inductive inference? Give example(s) and solution(s)

Answer 1

What is meant by the term overfitting in the context of inductive inference :-

• Inductive inference is the process to reach on general conclusion from specific example.
• The main advantage of viewing inductive inference system of inductive bias is that it provides non procedural means of characterizing the policy for generalizing beyond the observed data.
• And it also aloows for comparision of various learners based on the strength of inductive bias.
• The following algorithm can be based on weak and strong bias --
• Elimination algorithm for candidate -- New instances can be classified where all the mambers of current version space agree in classification.
• Inductive bias -- The target concept can be represented in hypothysis space.

Example --

• For example there is some relevent attribute x,y, z.
• The valus of x = 2 , y = 3, z = 5.
• The ovrfitting of x, y, z can be
• Solution --  Model 1 -- x + y = z, here the prediction can be x = 0, y = 0, z = 0
• Model 2 -- If x =2 , z = 5, then y = 3, otherwise y = 1
• Good -- Fully consistent with data.
• Bad -- Justification can not be given in the data for the prediction of y = 1 in all cases.
• Inductive bias -- The assumption can be based on implicit and explicit for the kind of model is wanted.