Prospect and issues in Machine Learning

Prospect and issues in Machine Learning

Perspectives of machine learning

Perspectives of machine learning searching a very large space of possible hypothesis ( it can be a explanation for anything) to determine one that best fits the observed data and any prior knowledge help by learner

Example

• It represented algorithms

Choosing a merge sort algorithms is the perspective based on the required peek up algorithms.

Issues in Machine Learning

What algorithms should be used because you have n number of algorithms, decision tree algorithms, inductive biase, basian algorithms etc,. Machine learning has so many algorithms among which algorithms choose it is one of the big issue.

Which algorithms perform best for which type of problem

You have different types of problems like checkers problem, hand written Recognition problem etc,. What type of problem which algorithms used to give a best results.

How much training data is sufficient? And testing data.

What kind of methods should be used? Which type of algorithms, what methods should be used.

What methods should be used to learning overhead in order to reduce the learning overhead. Machine what kind of methods should be used like directly training a machine

For which type of data method should be used?