
A.I.de
This app prototype was created to demonstrate an app that connects machine learning researchers with large samples of users to increase data collection and accuracy.
"Create something to improve machine learning and artificial intelligence in robotics" is a hefty request in general, and even more so when you have less than 48 hours to do so! Regardless, that is the problem my team and I tackled for the 2020 WEHack hackathon.
A.I.de (which comes from mixing A.I and aide) is an app that acts like the popular mind puzzle apps on the market that give to a set of problems, questions, or riddles for a user to answer, but in this case, those problems and questions are from research groups looking to collect data to train their algorithms with.
We all know that a lot of the tech for amazing machine learning programs already exists, but an area where researchers are constrained is finding data to train it with, which is where our idea began.
To tackle creating this app, I created flow charts for the "user side" of the app as well as the "researcher side" of the app. The user would be able to open the app, play games, and earn points that can be used to get real-life rewards and incentives. The researcher must create an account and go through a verification process before they start creating questions or games to collect data.
This prototype was designed using Arduino Studio and was coded primarily in Java to create a landing page, interface/game page, and rewards page.
I was elated that others enjoyed this idea as well and that my team won first place out of hundreds of entries from all over the country!
View the GitHub: http://bit.ly/AIDE_Github
View the Presentation: https://bit.ly/aideppt