Machine learning is a hot topic in computer science right now. A large number of case show that machine are able to learn and make prediction accurately. Machine learning has a strong connection with statistics, maths and computer science. It enables people to analyze large number of data.
The easiest method in machine learning is linear regression. Linear regression is the basis of other kind of regression like polynomial regression. Linear regression are powerful enough to give you insight about data set.
Based on linear regression, we can add a sigmoid function to convert it to logistics regression. logistics regression helps the machine to understand the logic.
Based on logistics, people developed neural network, more specifically, multi-layer perceptrons. Every perceptron is a logistics regression unit. Its input is the previous unit.
Another powerful tool is Support Vector Machine. Similar to regression, its cost function is based on the cosine value of the vector. What’s more, this method provides strong mathematical explanation of the prediction model rather than the pure black box of neural network.