Start Here:
- AutoML.NFT Dapp is running live on the cloud.
- Sample "restaurant" dataset can be used to test the dapp. Under role, role for "RestaurantID" should be "Id" and role for "Satisfaction" should be "Target".
- Source Code for the project is available under github.
- Documentation describes the project and the dapp.
- YouTube Video is the demo video for the hackathon.
Executive Summary
- a Machine Learning as a Service (MLaaS) platform to
- interactively and dynamically generate Machine Learning (ML) report for uploaded data files, and
- issue NFT tokens on the XRP Ledger as certificate of ownership.
AutoML.NFT also provides an analytics marketplace, where legal ownership of ML reports can be put on sale and bought. The project also provides a template for XRPL and other blockchains on how MLaaS dapps with innovative features can be developed on the blockchain.
Inspiration
Machine learning as a service can enable end-users in business, industry, governent and NGO, to run automated machine learning (AutoML) easily. To this end, we built -to the best of our knowledge- the first ML as a service platform that accepts crypto payments and issues NFT certificate of ownership. The system also contains a marketplace where data analytics reports can be put on sale and be bought.
What it does
Our project is a ML as Service (MLaaS) system running on the blockchain. The system runs on Web 3.0, records contracts and other data openly on the XRP Ledger (XRPL), and issues certificate of ownership of generated analytics reports as NFT. The project integrates multiple blockchain project categories, namely NFT, ML, marketplace, DataFi/analytics, under a single operational framework, and implements this novel framework as a working dapp.
How we built it
Python for ML and accessing XRP's API for smart contracts. Python ML libraries include Pandas, seaborn, matplotlib, sklearn. Front end is implemented using Javascript, React, Node.js, and XRPL.
Challenges we ran into
Using a browser-based wallet (such as Metamask) for XRP was not possible instead we managed wallets with the help xrpl.js library. Integrating the ML code was also a significant challenge.
Accomplishments that we're proud of
A system that integrates smart contracts, marketplace, ML engine, and other features in a single working dapp.
What we learned
Most importantly, we learned that this can be done. Secondly, we learned that we still need to discover the MetaMask of XRPL if there is one, or that it would be great if developed by Ripple or the community.
What's next
So much planned. Check our our github project's Documentation page, under "Future Plans" section.
Built With
- javascript
- matplotlib
- node.js
- pandas
- python
- react
- seaborn
- sklearn
- xrpl
Log in or sign up for Devpost to join the conversation.