Sometimes the acronyms in AI can feel like you're getting snowed in...We're here to help
What is Federated Machine Learning (FL) and how does it differ from other types of Machine Learning?
Federated machine learning (FL) brings the ML model to the data source rather than moving the data to the model. This allows decentralized data sources to train the model without having to move the data to one location.
How will Devron fit into our existing ecosystem? Will it require heavy lifting on our end?
Synergy sits as an API layer on top of your federated cloud data lakes all connecting to a central unified environment where a Data Scientist can develop a model, without centralizing data.
Will Devron have access to any sensitive information?
Devron never has access to your sensitive data. We provide our containerized software stack to you with a deployment strategy to help you create your federated analytics environment. What we do manage for you is your Synergy deployment inside your enterprise which would be a set of containers (using Docker) that we can manage per vertical use case, again with no access to your data.
What layer of data security does Devron provide?
Our platform has been tried and tested by the U.S Intelligence Community (specifically the NSA and the CIA). We provide enterprise-grade security, authentication controls based on your active directory and policies - as well as layers of encryption on the machine learning models so they can not be reverse-engineered and leak any sensitive data.
What customization can we request?
Synergy offers various machine learning algorithms off the shelf that are ready for federated learning from day one. Customers can request other specific algorithms that you’d like us to create for your federated learning architecture. For example, if you’re interested in some deep learning model architectures then we can gladly work with your team to create them for encrypted federated learning. Feel free to contact us at email@example.com to explore our customization capabilities.
What tools can Synergy's Aventail SDK integrate with?
Synergy primarily uses AWS services, with API ergonomics with Tensorflow, PyTorch, and Kubernetes as well as warehouse and big data technologies such as Spark, Hadoop, Snowflake, and Palantir.
How long will deployment & implementation take from start to finish?
From the point of software installment and deployment to your internal enterprise cloud - a data scientist can develop an encrypted federated learning model in just hours on your enterprise federated data without needing to centralize it.
How involved will Devron be during deployment and implimentation?
Devron sends a forward deployed engineer to make sure best practices are being followed in your enterprise to develop, test and deploy a federated machine learning model based on your privacy budget. A privacy budget is simply how much privacy of your data needs to be preserved at the cost of model accuracy.
Who from my organization can access the models and can we add user access controls?
Primary users of Synergy will be data scientists that are tasked to create analytical models from sensitive datasets. As teams are being created and multiple projects and experiments are being run, you have options to add additional users.
General Data Protection Regulation (GDPR)
California Compliance Privacy Act (CCPA)
California Privacy Regulation Act (CPRA)
Effective: January 2023