DISCO is our [DIS]tributed [CO]llaborative Learning platform!
A code-free, installation-free browser-based platform that allows any non-technical user to collaboratively train AI models without sharing any private data.
---
š¤Share Models Not Data: Jointly train AI models with collaborators without sharing data. [Private data is never shared or uploaded online] āØDecentralized andĀ FederatedĀ learning supported
Ā šĀ Accessible open-source, open-access: 100% in-browser learning with no installation steps and no coding skills required. Works on any device [š»š±š¤³š½š„ļø]
šĀ PlugnPlay: To join aĀ DISCOllaborative simply 1) select a model and 2) hit the 'Train' button
š¤¹Versatile: support language models, disease classification, image recognition, multi-class classification, and others
š§Customizable:Ā Deploy your own DISCO instance now. Code is purely in JavaScript/TypeScript [Can be run in-browser or in node.js ]
šļø#WatchThisSpace Many more features to come!
---Ā
Explore the working demo herešhttps://discolab.ai/#/
---
Code:Ā https://lnkd.in/epp3EEySĀ
Technical release notes:Ā https://lnkd.in/eM8pf7Kn
Give us feedback:Ā https://lnkd.in/eS5bWy9N
Made with lā¤ļøve, byĀ EPFLĀ Yale School of MedicineĀ EPFL School of Computer and Communication SciencesĀ and Yale Biomedical Informatics & Data ScienceĀ
Martin JaggiĀ ValĆ©rian RoussetĀ Julien VignoudĀ Walid BennaceurĀ Klavdiia NaumovaĀ Eduard ÄurechĀ Tahseen Rabbani
Thank you,Ā International Committee of the Red Cross - ICRCĀ Javier Elkin, PhDĀ Blaise RobertĀ Fabrice LauperĀ Michael DayanĀ Melissa Harper Shehadeh.Ā We love making useful things with you!