Model Training
There are many different machine learning models for each of those four types of learning algorithms. Machine learning models rely on some forms of mathematical/statistical models. When we train models, it means we use an algorithm to find out the model's unknown parameters. Scientifically speaking, we cannot definitively find out the ground truth for unknown parameters. Instead, we can only estimate the unknown parameters as closely as possible to the ground truth by using mathematical/statistical methods on sample data. Estimating unknown model parameters is equivalent to solving a mathematical equation whose solution comes in one of two forms: closed or non-closed.
Closed-Form Solution
Some algorithms' mathematical models have closed-form solutions. A model with a closed-form solution can be solved by expressing the model parameters analytically in terms of a finite number of certain "well-known" functions. A classic example is a linear regression...
