A trip recommendation app, based on neural network technology, written in Rust and Rocket. Locations are currently hardcoded to locations in Melbourne CBD and Clayton. Developed at UniHack Mini 2018, Monash University.
The latest nightly version of Rust and Rocket is required for the app to compile and operate properly.
Clone the repo, then
cargo buildat the root folder.
The neural model can be generated by
cargo run --bin genwhich will generate an initial, usable neural model from scratch.
Each generated neural model will be stored at the neural_networks/ folder.
Ensure that results.rs and training.rs points to the correct neural model filename.
The application can be served by running
cargo run --bin rocketThe app will be served at http://localhost:8000.
WARNING: Make sure that there are (at least) two model files at the
neural_networks/folder before doing the training mode. Also ensure that theneural_training_completionfunction points to correct model paths.
After running the application server, the training mode can be initiated at http://localhost:8000/training.