EyA Hones in on Federated Machine Learning using Conclave from R3

2021-12-01

With the EyA platform becoming a standard for many businesses, it was time to begin forging the path toward V2 of the enterprise cloud. A vast number of new features are going to be bundled into the next release, but with the world desperate for a data intelligence solution, EyA picked up the batton.

Currently data lakes are used in the main to perform complex analytics on data sets harvested from one to many sources. The process from harvesting to actually initiating the analysis is a very long, costly and inefficient process. Even prior to the data collection, vast quantities of time and data scientist work is spent in calculating the worthiness of the data to be analysed.

Many data lakes have failed because they were IT-led vanity projects, with no clear linkage to business objectives and operational processes. ... Failed data lakes often represent a toxic combination of both poor technology choices and an inadequate approach to data management and integration.

Increasingly, the generators and consumers of data are intelligent systems – connected infrastructure, smart cities, and IoT edge analytics and AI. We have moved on from the consumption and curation of information for a 8 billion people to curation and consumption of data by trillions of devices.

 As the volume of data increases and the consumers of that data is intelligent systems, the need for automated data provenance has expanded. Not only that, we need to democratise data through decentralisation and have an appropriate model for its currency. This is the problem that we need to solve – at scale, at speed, and in all aspects related to trust.

EyA is an entire stack of technologies built on permissioned based DLT, which is much more than just a blockchain. Taking Corda Enterprise from R3, the solution has been developed beyond the completely private node system of Corda, into a hybrid inter-private node of nodes.  The enabler of complex data relationships is the semantics templating engine, which actually mimics a relational database solution within the DLT.  With this is mind, dynamic and complex relationships between any form of data are organically developed without human intervention.

Organisations are also able to mark data at birth with varying levels of permissions and access control.  When multiple organisations opt into a contract to loan / rent data, no data can ever be seen by the renting party and all analytics are performed within Intel’s trusted computing technology removing the barrier of trustless computing.  With direct bias toward the semantics of anything from a sub atomic particle through to complex structures, people and cities or countries, classification of data is governed throughout the lifecycle of any given entity.