The Minimum Description Length Principle in Coding and Form

Instructions and help about The Minimum Description Length Principle In Coding And

in this video as part of bayesian learning we will study a minimum description length principle minimum description length shortly called as mdl principle involves a way of trading of hypothesis complexity for the number of errors committed by the hypothesis we are always interested to have less hypothesis complexity the thing is when you go for lesser complexity hypothesis we may have to accept with more number of errors when you have high complexity of hypotheses we may have less number of errors so it might select a shorter hypothesis that makes a few errors over a longer hypothesis that perfectly classifies or classifies to a perfectly classifies or classifies to a nice layout classifier is nicely right the training data viewed in this slide it provides one method for dealing with the issue of overfitting the data one way of solving over fitting of the data using this mdl principle i believe you know what is overfitting of the data overfitting of the data means if you do not know p