Each email has a label 0 or 1, either spam or not spam. Let us say that we want to classify spam emails, probably the most often used example in machine learning (note, this is not a post on naive Bayes). Let's start with a simple supervised learning classification problem. ![]() This article is heavily based on the book Understanding Machine Learning by Schwartz and Ben-David, which I highly recommend for anyone interested in the fundamentals of learning theory. I will try to explain the underlying concepts as simple, short and theoretically grounded as possible. In my opinion, anybody that is serious about machine learning should be comfortable with talking about ERM. The theory behind ERM is the theory that explains the VC-dimension, Probably Approximately Correct (PAC) Learning and other fundamental concepts. Understanding ERM is essential to understanding the limits of machine learning algorithms and to form a good basis for practical problem-solving skills. Empirical Risk Minimization is a fundamental concept in machine learning, yet surprisingly many practitioners are not familiar with it.
0 Comments
Leave a Reply. |