OpenMined/PySyft
PySyft
PySyft is a privacy-preserving machine learning framework and remote computation engine. It functions as a decentralized data analysis orchestrator that allows for the execution of data science workflows on remote servers without requiring the transfer of raw private data from the host device.
The platform provides a secure collaboration environment where data owners manage permissions and authorize specific collaborators to run computations. It differentiates its workflow by utilizing mock data for local development and validation before submitting final analysis jobs to private remote servers.
The system covers a broad range of secure computation capabilities, including the use of sandboxed job execution to isolate computations from the underlying system and a cloud-storage transport layer for exchanging requests between peers. It also includes mechanisms for asynchronous state synchronization to maintain consistency across offline or cloud-connected nodes.
Features
- Remote Data Analysis - Executes machine learning workflows on datasets hosted on remote servers to maintain strict data privacy.
- Privacy-Preserving Machine Learning - Implements a framework for executing machine learning workflows on remote servers without accessing raw private data.
- Privacy-Preserving Compute Engines - Provides a remote computation engine that allows analysis jobs to run on private data sources without raw information leaving the host.
- Decentralized Analysis Orchestrators - Provides a decentralized orchestrator for coordinating remote analysis and synchronizing computation requests across offline or cloud-connected peers.
- Decentralized Data Analysis - Orchestrates the analysis of information spread across multiple offline or cloud-connected servers.
- Remote Function Execution - Sends computation logic to the data source, ensuring raw private data never leaves the host device.
- Privacy Orchestration - Coordinates remote data analysis and synchronizes computation requests across distributed peers while maintaining data privacy.
- Data and Resource Permissions - Provides granular controls for restricting visibility and operations on private datasets by requiring owner approval.
- Role-Based Access Control - Requires data owners to manually approve specific collaborators and computation actions through role-based permissions.
- Secure Data Collaboration - Manages permissions and approvals for multiple researchers to work on private datasets hosted by different owners.
- Privacy-Preserving Workflows - Implements a workflow cycle of using mock data for development and remote execution for production results.
- Secure Permission Management - Offers a platform for managing data permissions and authorizing collaborators to run computations on isolated private datasets.
- Mock Data Utilities - Provides synthetic data surrogates to validate analysis code locally before deploying to private remote servers.
- Code Execution Sandboxes - Runs remote computations within isolated virtual environments to prevent unauthorized access to the host system.
- Isolated Execution Sandboxes - Runs submitted computations in sandboxed virtual environments to restrict access to the underlying system.
- Federated Learning - Library for secure and private deep learning.
- General Machine Learning - Library for secure and private deep learning.
- Guardrails and AI Safety - Listed in the “Guardrails and AI Safety” section of the The Incredible Pytorch awesome list.
- Privacy and Safety - Library for secure, private deep learning using MPC.
