Some Basic Machine Learning Terms #1

If you are working on some project related to machine learning (ML) or you are a newbie researcher knowing these terms and definitions might be useful.

Machine Learning: A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E — Tom M. Mitchell

-> There are two main types of learning process (some more s.t. reinforcment learning, transduction)

Supervised Learning: is a way of learning some fact by using given features of prior instances. As another saying, you have a data-set including instances, their features and outputs for you to predict next coming instance’s output.  For example you might have a list of (X,Y) couple for each instance and X=size of the house and Y=price of the house. Then you have a new instance having (X=1200m^2, Y = ?) and you try to predict Y by considering already defined instances and their X and Y values. For more info…

Unsupervised Learning: is more like clustering, grouping some instances other than getting a certain answer. You have a number of instances and by looking their features, attributes you try to group them together. You might see the example implementation from Senior project Bunch-It.com that collects search engine results and group them together under a label or Google news web page that collects all news articles under a topic or event. For more info…

-> General ML subjects are based on supervised learning and there are two main process on supervised learning.

Classification: is learning process on discrete domains. It selects one of the answer from the discrete domain. It might be colour {red, blue, green} of the thing, size {S,M,L,XL} o such. For more information…

Regression: is estimating a values by considering given features of the instance but this estimation is based on continues domain like real number. For example estimating a value of a stock price by considering some attributes is a Regression process since the price is inside a continues real number domain. For more information…

This is an initial explanations about  ML and that’s all for now. If you are interested in ML I will be posting more. Keep watching 🙂

Share