Some people learn quickly; some people learn slower than others. I'm a firm believer that anyone can learn anything.
The hardest part of learning (often) is understanding something that's new and different and isn't accounted for by any of the stuff that you've managed to pack into your head.
Maybe you've experienced this: "X wasn't doing well, until they met Y, who was a brilliant mentor and guided X."
Hopefully that sounds familiar; personalized attention from someone who deeply understands a concept and also understands how you learn can make the difference between you learning, or not.
What if I told you that you could have that personalized attention? What if I told you that it could be free?
Everyone learns a little bit differently, and I'm saying that anyone can learn anything and we can help make that happen.
The way we're going to do that is:
- Build a dataset of educational content (lots of these exist.)
- Produce an embedding space for this dataset.
- Use a tree based algorithm from this space to produce content samples for students learning about particular topics.
- Measure those students.
- Use a learning algorithm to guide back-propagation in this embedding space as a function of the measurements.
- Repeat.
The learning agent is deciding how best to nudge the embedding vectors to maximize performance of the student learners.
So, what's done, what needs to be done, and what's the status?
This is a Python package which can construct a KD tree and associated embedding space for a dataset representing a collection of educational content.
I just started this; considering my availability I'm expecting to see this done, with implemented testing, sometime late 2021 when it's done.
This wraps the kit and kaboodle above in an OpenAI API compatible learning environment that it also compatible with RLLib.
This will be done after the information tree construct is available for use.
These experiments, using machine learning constructs as learners in a controlled environment, are intended to serve as proof of concept.
You know what, yes! I would love a project manager; if you're interested in the technical aspects there's certainly more room for hands to be put on.
Questions are welcome.