Curve SVM with AI. Hyperbola SVM? Can we build such SVM models with AI?

Here is what I wanted: to separate two classes using a 2D curve, then map it to higher dimensions. Let’s start by separating with a parabolic curve. I am building this model with AI as collobrator.

Here is the first prompt:

“Just like SVM is a model based on the separation of two classes of data by 2 planes, in the same way, generate a model that separates two classes of data by optimal curves, maybe two classes separated by a parabola (hyperbola) to start with.”

Here is a summary of ChatGPT’s response.

Second prompt to ChatGPT:

“Visualize hyperbola separation of two classes of data in 2D.”

ChatGPT response:

Here is hyperbola constraints,

What I’ve built (big picture)

This model is essentially:

• A quadratic classifier
• But explicitly interpretable as a hyperbola

Map into higher dimension WITHOUT flattening

Here is how Claude summarized it:

Here is a summary from Meta AI, Meta connects it to the Bayes boundary region we studied in Machine Learning, here,

ChatGPT gave the best part.

Lets build this model, Curve SVM, as ChatGPT named it!

Thank you for reading.

I ll update when I get to final version of what I am building with AI.

Subscribe for updates.

Regards,

….