How to use Quantum Multi Agent Reinforcement Learning for time-series prediction

PennyLane Help
1

Hello, I am competing in a science fair and my project heavily depends on using Multi-Agent Reinforcement Learning supported by VQC’s to make time-series predictions, like the stock market. There is a github that implements MARL with a VQC here: GitHub - WonJoon-Yun/Quantum-Multi-Agent-Reinforcement-Learning: Quantum Multi-agent Reinforcement Learning (QMARL)
but I don’t know how to edit it for a more timer series perspective. In the example of stocks, how would you create agents that would evolve to make time-series predictions given factors like closing price? I am still learning ML and QC so any help would be appreciated. Thank you!

2

Hi @Aadi_Tiwari, it’s great that you’re learning Machine Learning and Quantum Computing!

In QHack 2022 there was a finance challenge where teams could look for a finance problem to solve with quantum computing. You can find those teams’ projects here.

You can find the teams who won the top spots at the end of our QHack 2022 blog post.

I encourage you to take a look at their projects and get some inspiration on how they solved their problem!

Also, if you would like to join QHack 2023 yourself you can go to https://qhack.ai/ and register starting on December 5th.

I hope this is helpful and please let me know if you have any further questions.

3

The QHack 2022 posts were quite impressive and there were a few posts that were similar in idea, but right now I’m struggling to even find a classical implementation of Multi-Agent Reinforcement Learning for Stock prediction. I don’t think there would be a QMARL for time series prediction then lol

4

That’s true @Aadi_Tiwari.

I did a quick search on the arXiv and found a couple of papers on Quantum Multi-Agent Reinforcement Learning. They don’t do exactly what you’re looking for but they could be a good starting point if this is something you want to implement yourself.

A lot of things are yet to be implemented and this is exciting!

If you do make an implementation yourself it would be awesome if you could share it here, or as a demo, so that others can benefit from it too.