Having a passion for striving to combat educational inequality, we had the intention of making a free, online tool that will help elementary school students with arithmetic. Being fans of machine learning, we wanted to integrate neural networks into this project (in the form of brains that the child trains by answering questions correctly).
What it does
The user answers various arithmetic questions. Based on correctness, their virtual brain gets better or worse. Their goal is to have the virtual brain answer questions as correctly as possible (which is only possible if the user answers their set of questions correctly as well).
How we built it
We used Keras for the neural networks. Then, we used Flask along with Bootstrap on the front end to make the web application. We used an SQL database for storing user data.
Challenges we ran into
We had many issues setting up the GitHub repository and connecting to our editors for easy pushing. We also were very ambitious about learning new technologies so we spent more than half of our time trying to learn React, which we dropped and did not end up using in the end.
Accomplishments that we're proud of
We are proud of the "meat" of our project. The neural networks work really well and train pretty well if the user answers their questions correctly.
What we learned
We learned a lot about managing project revisions. Also, we learned a lot about planning the project from the beginning, which would have saved us a lot of time had we done that from the start.
What's next for Big Brain Battle
Our next step is to add a feature where each user will be able to compete with other users. It will be a 2v2. The user will compete alongside with their trained virtual brain.