Introducing the Rhino Federated Computing Platform
The Rhino Federated Computing Platform (“FCP”) provides privacy-preserving edge computing and federated learning as-a-Platform, with robust support for the Machine Learning Operations (MLOps) workflow. In essence, we let AI developers work with data without ever taking possession of the data. Using the Rhino FCP, AI innovators can easily curate and annotate data, as well as train and deploy machine-learning models across multiple institutions or with industry partners, without any need for sharing data.
.png)
Key features of the Rhino FCP include
Data validation - Auto-identification of data, generating reusable templates (‘schemas’) that can be used for validation and versioning
Generalized Compute (GC) - Run any custom code on data, ‘on the edge’
iGC - Interactive with a broad range of softwares (including GUI), without the need for lengthy integrations
Secure Access - Integrated zero-footprint data and image viewer / annotation tool
Analytics Toolbox - exploratory data analytics, and a range of epidemiological and statistical analyses, including results visualization
GUI / SDK - Access via a web-interface and a programmatic interface
Integrated Federated Learning - Integrated with NVIDIA's FLARE™ framework with enhanced debugging and real time training monitoring & control with TensorBoard™
Security & Privacy
Maintaining our partners’ trust is paramount, and so the FCP is built on the principles of privacy by design, with a best-in-class security posture. Key security & privacy features include:
Data always stay at rest behind custodian firewalls, with no direct server access, and no inbound ports. Edge computing and federated learning mean that all processing is done locally, with only summary statistics or model weights leaving the site
Secure container architecture means that code run on local data cannot access the internet to exfiltrate sensitive data
Data are encrypted at rest and in transit
Role-based Permissions allow data custodians to control exactly who has what access to data for what purposes and for how long
The FCP logs all user actions, available via visual log
De-identification: The FCP has pre-built HIPAA safe harbor logic, as well as the ability to run custom de-identification software. The FCP also enables k-anonymization and differential privacy
Rhino Health is ISO 27001 and SOC 2 certified, and has received third party assessments of compliance with HIPAA and GDPR