DISCO: A Browser-Based Privacy-Preserving Framework for Distributed Collaborative Learning

2025-11-24

AI Summary

Data silos—arising from privacy regulations, legal constraints, and intellectual property rights—undermine statistical power and exacerbate accessibility bias in machine learning. To address this, we propose a lightweight, browser-based distributed collaborative learning framework enabling non-technical users to jointly train models without sharing raw data. Our modular architecture unifies federated and decentralized paradigms, integrating client-side browser training, a frontend ML inference engine, end-to-end encrypted communication, and multi-tiered privacy safeguards. It supports customizable weight aggregation strategies to enhance model personalization and robustness against bias. The open-source platform is cross-device compatible—including smartphones—and requires only a web browser for participation. Empirical evaluation demonstrates significant improvements in usability, fairness, and scalability of collaborative modeling while preserving data confidentiality and regulatory compliance.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Computer Vision: Bias, Fairness & Privacy

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Data is often impractical to share for a range of well considered reasons, such as concerns over privacy, intellectual property, and legal constraints. This not only fragments the statistical power of predictive models, but creates an accessibility bias, where accuracy becomes inequitably distributed to those who have the resources to overcome these concerns. We present DISCO: an open-source DIStributed COllaborative learning platform accessible to non-technical users, offering a means to collaboratively build machine learning models without sharing any original data or requiring any programming knowledge. DISCO's web application trains models locally directly in the browser, making our tool cross-platform out-of-the-box, including smartphones. The modular design of disco offers choices between federated and decentralized paradigms, various levels of privacy guarantees and several approaches to weight aggregation strategies that allow for model personalization and bias resilience in the collaborative training. Code repository is available at https://github.com/epfml/disco and a showcase web interface at https://discolab.ai

Problem

• Enables collaborative ML without sharing original data
• Provides privacy-preserving distributed learning for non-technical users
• Offers federated/decentralized training with model personalization options

Innovation

• Browser-based federated learning without sharing data
• Cross-platform local training including smartphones
• Modular design supporting privacy and personalization