How Feature Transformation Makes Linear Classification Easier?
Feature Transformation is key player in Machine Learning algorithms as its the one which transforms abstract data into vectors. In perceptron algorithm , generally if we see the case where linear classifier or the hyperplane doesn't pass through origin , we can apply feature transformation to make linear classifier pass through origin.
Basically what feature transformation does is that it adds another dimension to our data and as result of that it redefines our hyperplane equation in such way that it can make our linear classification easier , if not through origin then will accurately classify the data.
Get Muhammad Talha Fayyaz’s stories in your inbox
Join Medium for free to get updates from this writer.
We get new equation for our vector as it passes through feature transformation. And the equation of hyperplane adjusts accordingly !
This is of great importance in multi-modal neural networks design and kernel networks designing.
That's what I studied today. You can share your thoughts at this.
You can connect with me at LinkedIn.
