Redux

tailwindcss

NextJS

React

D3

This web application was built to supplement the paper 'Predictive Design of Sustainable Biobased Packaging via Machine Intelligence for Improved Postharvest Preservation' as an open-access platform when community members may interact with our experimental datasets, and the ML-based tools we have developed in order to foster collaboration, and accelerate progress.

This project contains two main features: Forward Prediction and Inverse Design.

Perform direct single-point inference on the model based on the input composition/formulation specified.

Large synthetic datasets can be constructed via the trained predictive and/or generative model. Put simply, this is achieved by sampling input features from a gaussian distribution, while satisfying any constraints (ie. material compositions must sum to 100%). Output labels are generated via the trained predictive or generative model. This dataset can be queried to select for inputs that yield coveted output labels.