Machine Learning Paradigms
Overview of 3 most important machine learning paradigms: Supervised Learning, Unsupervised Learning, Reinforcement Learning
There are many machine learning paradigms. The three most important paradigms are:
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
The learning strategies in these 3 paradigms are very different and I will go through them one by one in the following section.
Supervised Learning
As shown in its name supervised learning means learning with supervision. As machine learning is aimed to learn a model from the data, so for different machine learning paradigms, the data presented to the machine is different. So, in supervised learning, a set of data are presented to a machine, and each data instance is a pair of input data instance, and the desired output is also known as labeled data. The goal of supervised learning is to try to learn a function or a predicted model from input to output.
In mathematics: Given a set of {xᵢ, yᵢ} for i = 1….N, where xᵢ is an m-dimensional numerical vector, and yᵢ is a scalar. During the training, the goal is to learn a mapping or function f: x → y by satisfying…
