Introduction to Machine Learning.

Introduction

• A computer program is said to learn from experience E with respect to some task T and some performance measure P. If its performance on T, as measured by P, improves with experience E. — Tom Michell.
• Machine learning is classified into 4 types of learnings: Supervised learning, Unsupervised learning, Reinforcement learning, and Recommendor systems.

Supervised Learning

• This learning is again classified into 2 types: Regression and Classification model.
• Regression–goal to continuous valued output.

Classification

goal to Discrete valued output like 0,1,2,3…for the sake of understanding it uses only 1 feature, but there are some algorithms where we use an infinite number of features as a X-axis. Moreover, it might be hard to handle that huge number of features and we may run out of memory, at that time we use SVM (Support vector Machine) to handle such situations.

To examplify it, consider the below image, like based on the tumor size we are going to classify it whether it is Benign or Malignant!!!!

In the first pic only 1 feature is used and in the 2-pic it uses 2 features. In the similar way we can use an infinite number of features while we make predictions.

Unsupervised Learning

• This learning also called Clustering mode, where we will give data and tell the algorithm could you find the structure in the data or can you automatically cluster the individuals into different types that I don’t know in advance. The best way to exemplify this model is “Cocktail Party Problem”.

The cocktail party effect is the phenomenon of the brain’s ability to focus one’s auditory attention on a particular stimulus while filtering out a range of other stimuli, such as when a partygoer can focus on a single conversation in a noisy room

But here the model itself clusters the data, upon assigning as a input to the model.