I am trying to explain the geometrical intuition behind neural networks. For this I am using a 2 feature input and a linear classifier as my theoretical elements.
What I wish for is to be able to manually add points of color, first on the plane z=0, then on the plane z=1, then plot a vector (an arrow basically) whose tail is at the origin and the associated normal plane, and finally I want to take the 2 sections of the data plane (i.e when the plane defined by the vector intersects the data plane, it will divide it in 2 sections) and color them with different gradients. By the last I mean, I first want to color the data plane with a linearly interpolated color to denote a linear gradient that increases with the slope of the intersecting plane and then I want to color it with a gradient that is the square of the difference of the distance of a point in the data plane to the intersection line.
I am sorry if the explanation sounds confusing, it makes sense for me as shapes, and I am not sure how to describe it other than that (I am also heavily sleep deprived as I am writing this).
In other words I need some software that can plot 3D shapes, such as Geogebra or Desmos, but that also gives me the freedom to edit things with colors and "drawings". So imagine that MS paint and Geogebra have a child. I want to kidnap that child.
Is anyone aware of something that could be able to do this?