The concept is to classify the data with self-improved algorithm. By adjusting the separation line function, it is possible to approach the exact line correctly classifying the labeled data.
Potential Problem The model couldn't solve non linear separable data, as the model is constructed with linear functions
In the training process, the colors of background and symbol are designed as blue for group 1 and red for group -1. Simplified data and problem can well demonstrate the algorithm.
The figure shows an example of non linear separable data, where some points in -1 group are located inside the range of 1 group.
