ML algorithms recommand of online/batch learning for classification, prediction( and targetfunction), dataset parameter and label (A, B, C, Label)

Currently i am in a project. I will receive processing data constantly online from CNC machine, which will be like a dataset with parameters and labels, for example [A,B,C,Label],like 1st picture. The points(A,B,C) will be clasificated according to label. The 3 dimension classification surface would be like in the 2nd picture, above the surface labels of points are 1, below the surface labels of points are -1.

What i need to do are:

1. Find appropriate Online/ Batch ML algorithms to update the model. Of course also plotting the desicion boundary in 3 dimension when new data collected and new model trained(like 2nd picture).→→→ any suggestions about ML algorithms and plotting methods?
2. There will be a Target function F(A,B,C) to maximize (Parameters with constraints, min/max sort of), basicly it would be like find the point which meet F(A,B,C) = constant * A* B * C = MAX, but the Point(A,B,C) need to be under the desicion boundary, which means the label has to be -1.→→→ any suggestions about ML algorithms or other methods?

Data example

Desicion boundary