privacy-tech-lab / privacy-pioneer-machine-learning Goto Github PK

We should have a place somewhere (can be this issue or somewhere) else where we discuss parameter tuning, intermediary results, post screenshots related to that, etc. @danielgoldelman, I would leave it to you what you think is best.

Issue for understanding the training and testing process of ML models for textual binary classification.

We would like to experiment with incorporating ML into Privacy Pioneer. This issue will be for experimentation with loading and packaging Python code for use in the JS environment we use in Privacy Pioneer.

It is a good time for this task now. You are in the best position for this issue, @danielgoldelman.

@danielgoldelman will explain the above (and whatever else is relevant) in the readme of this repo.

(Here is the overview that @harkous created a few months back; if that is still how we do it, let's include a link.)

As we now have some data fully annotated, we must put it into this repo to begin the ML.

It seems the organization of the ML scripts is a bit convoluted at this point. Let's clean that up. At the latest, we should take some time at the end to organize. But if we are in a position to do it earlier, it will help us to approach the ML more systematically. So, the bottom line is to do it as early as possible.