You will find there's shut relationship amongst machine learning and compression. A method that predicts the posterior probabilities of the sequence provided its entire heritage can be utilized for optimum details compression (by using arithmetic coding over the output distribution).Semi-supervised learning can solve the problem of not getting enou