Support Vector Machines


De DataFranca

en construction


Définition

...


Français

...


Anglais

Support Vector Machines

Support vector machines are very powerful algorithms built on an extremely simple premise. Pick the two examples of your different classes that are closest to each other. In practice, this is typically done by finding the Euclidean distance between every training example. These two examples are your support vectors.

Draw a line through the space between the support vectors. This is your hyperplane.

Wiggle the hyperplane around until it maximizes the total distance between the plane and each support vector. This is your cushion.

https://opendatascience.com/machine-learning-for-beginners/