What's all the buzz about swarm learning?
Data, we all know, is valuable—at least when companies can put it to good use. The roadblocks to getting the most out of data today are significant: proprietary data that enterprises don't want to share, government regulations restricting data sharing, and much more. One result is that research efforts are often duplicated and innovation stymied.
The concerns of sharing data, legal and otherwise, are real. That's where swarm learning comes in. More than just another decentralized data sharing mechanism, swarm learning allows for the insights generated from data to be shared without sharing the source data itself. The source data never leaves the control of the entity that has ownership and control of the data.
Because the data itself remains secure, your business retains the value of that data and you have not risked triggering any regulatory compliance problems. Still in its early stages of application and deployment, the basic principles of swarm learning and potential techniques for making use of it are becoming well understood; the trick will be finding ways to use these techniques in an effective fashion.
An important breakthrough? You bet. That's why we compiled the following articles to help you get up to speed on the subject:
Data is proliferating at the edge, but not all of it can be shared due to privacy or security reasons. Swarm learning offers a way forward, allowing you to share the insights and learnings from data without having to share the data itself. Dr. Eng Lim Goh, senior vice president and chief technologist for AI at Hewlett Packard Enterprise, and MIT Technology Review's Laurel Ruma recently sat down to discuss why swarm learning is more secure, and how it’s working to improve data analysis in sensitive environments, such as hospitals.
Researchers at Hewlett Packard Labs dive into swarm learning and how a distributed model can improve the use of machine learning and artificial intelligence in the analysis of the ever-growing mountain of data scattered across your enterprise.
Applying swarm learning techniques to decentralize data analysis and share data insights can accelerate medical research, long a challenge. With many regulations and data security requirements loosened to help the development of COVID-19 vaccines, the medical research field is beginning to understand the value of data sharing; swarm learning could allow a similar pace of research and development for other treatments, even with strict data security guidelines.
By sharing neural network parameters over blockchain, researchers can collaborate on data in different jurisdictions. "The beauty of swarm learning is that there is no central node which aggregates the data," says Dr. Eng Lim Goh, senior vice president and chief technology officer for artificial intelligence at Hewlett Packard Enterprise. "The swarm network acts as a union by sharing insights directly with all participants of the respective learning. There is no central custodian collecting all learnings or insights." This decentralized distributed model shares only the insights gleaned from the protected data and will enable a new era in research analytics.
As we constantly add more sensors, IoT devices, and data sources to our networking environments, we risk the tsunami of data overwhelming the network. Adopting swarm learning techniques at scale may be the only way to prevent your business from drowning.
As edge devices become more powerful, it becomes possible for data to be analyzed where it is collected. But determining the value of the data, and collating and utilizing that data as it is collected, still requires some form of analysis that factors in all the data being collected from all related sources. That's where swarm learning comes in, with the technique minimizing the amount of information that needs to circulate around the network.
Edge devices are the source of all our data. So where's it all going? We discussed the future with three leading voices in edge computing. Swarm computing is a major component that will allow businesses to get the greatest value as the edge moves to the center. Because the solution allows shared improvements in research and technology without ever giving up data sovereignty, use cases for swarm learning seem clear, especially when dealing with edge devices that continue to get more intelligent. It is, however, a significant change in the current approach to decentralized data analytics, one that will require both a certain degree of corporate will and, in many cases, changes in the way regulations concerning data privacy and sovereignty look at data that has been collected.
This article/content was written by the individual writer identified and does not necessarily reflect the view of Hewlett Packard Enterprise Company.