Inspiration
I was inspired by the idea of transfer learning. Using previous experience to transfer capabilities without complete retraining from the start.
What it does
This project demonstrates the possibility of using a machine learning neural network to make more data. Whilst the data is not perfect, it can be used to train a flood prediction model that is then fine tuned using real life data.
How we built it
This was built based initially on MATLAB's Neural Network Fitting Application and modified with scripts.
Challenges we ran into
The lack of data was super difficult to accept and led to this project challenge!
Accomplishmentsthat we're proud of
Successfully obtaining comparable results after the training, what an unexpected surprise.
What we learned.
There exists many problems to be solved with more problems being created everyday, there is no time to lose!
What's next for Yun
Sleep. Time zone differences are really bad for health when doing hackathons.
Log in or sign up for Devpost to join the conversation.