Training Machine Learning Models at the Edge
More and more data is being collected at locations far removed from the central core of the network. Some of this data can't easily be transferred back to the core for analysis, either because it is too large or because there are restrictions due to privacy or data sovereignty. This creates a challenge for Machine learning models, because nodes at the edge can drift if they don't remain in sync with changes to the model. Swarm learning helps address this by pushing the training to the edge near the data, with Blockchain providing a way to nominate a central host that aggregates the training and keeps everything in sync. In this interview from HPE Discover, Mark Linesch explains how machine learning at the edge works and some of the potential use cases.