Hewlett Packard Enterprise announces the April 28 launch of HPE Swarm Learning , a new AI solution that accelerates insights at the edge—from disease diagnosis to credit card fraud detection—by sharing and unifying learnings of AI models without compromising data privacy.
HPE Swarm Learning, which was developed by Hewlett Packard Labs, the R&D organization of HPE, is the first privacy-preserving decentralized machine learning solution for edge or distributed sites. The solution provides customers with containers that are easily integrated into AI models through the HPE Swarm API. Users can then immediately share AI mannequin learnings within their organization and externally with industry peers to improve training without sharing actual data.
“Distributed learning is a powerful new approach around AI that has progressed to solve global challenges, such as improving healthcare and optimizing anomaly detection, assisting in fraud detection and predictive maintenance efforts,” said Justin Hotard, executive vice president and general manager of HPC and AI at HPE. “HPE is contributing to the distributed learning movement in a significant way by delivering an enterprise solution that enables organizations to collaborate, innovate and accelerate the power of AI models, while preserving ethics, data privacy and governance standards of each association.
New focus on AI to harness insights safely at the edge
Today, most AI model training takes place in a central location that relies on centralized, fused data sets. However, this approach can be inefficient and costly because large volumes of data have to be moved to the same source. You may also be subject to data privacy and data ownership rules and regulations that restrict data sharing and movement, which can lead to inaccurate or biased models. By training models and leveraging insights at the edge, companies can make decisions faster, at the point of impact, leading to better experiences and outcomes. Also, by sharing learnings across organizations at the data source, industries around the world can unify and further enhance intelligence that can drive impressive business and social outcomes.
Notwithstanding, sharing data externally can present a challenge for organizations that must meet data governance, regulatory, or compliance requirements that require data to remain in its location. The HPE Swarm Learning Solution enables organizations to use distributed data at its source—increasing the size of training data sets—to develop machine learning models and learn equitably while preserving data privacy and governance . To ensure that only the learnings captured from the edge are shared, and not the data itself, HPE Swarm Learning uses blockchain technology to securely onboard members, dynamically elect a leader, and merge the parameters of the model for providing resiliency and security to the massive network. Additionally, because it only shares learnings, HPE Swarm Learning allows users to tap into large training data sets, without compromising privacy, and helps eliminate bias to increase mannequin accuracy.
“Massifying” data to empower AI for the common good
HPE Swarm Learning can help diverse organizations collaborate and improve insights:
- Hospitals can derive learnings from imaging records, CT and MRI scans, and gene expression data to be shared across hospitals to improve diagnosis of diseases and other ailments, as well as protect patient information.
- Banking and financial services can combat the expected global loss of more than $400 billion in credit card fraud over the next decade 2 by sharing fraud learnings with more than one financial institution at a time.
- Factories can benefit from predictive maintenance to memorize about and address equipment repair needs before they fail and cause unforeseen downtime. With mass learning, maintenance managers can gain better insights when they collect learnings from sensor data across multiple manufacturing plants.
Some examples of use cases from early adopters of HPE Swarm Learning are:
Aachen University studies histopathology to speed up diagnosis of colon cancer
A team of cancer researchers at RWTH Aachen University Teaching Hospital in Germany conducted a study to improve colon cancer diagnosis by applying AI in image processing to predict genetic alterations that cause cells to become cancerous.
The researchers trained the AI models with HPE Swarm Learning on three groups of patients from Ireland, Germany, and the United States and confirmed the predictive performance on two independent datasets from the UK using the same AI models based on massive learning. The results demonstrated that the original AI models, trained only on local data, outperformed using mass learning because the learnings, but not the patient data, were shared with other entities to improve predictions.
TigerGraph optimizes anomaly detection to help banks combat credit card fraud
TigerGraph , Provider of a Leading Graph Analytics Platform, Combines HPE Swarm Learning with its Data Analytics Offering on HPE ProLiant Servers with AMD EPYC Processors to quickly detect uncommon activity in credit card transactions. The combined solution increases accuracy when machine learning models are trained on large amounts of financial data from multiple banks and branches in different geological locations.
HPE Swarm Learning is now available in most countries. For more information, visit hpe.com/info/swarm-learning
HPE offers a total turnkey machine learning development solution
HPE also announced that it will remove barriers for enterprises to develop and train machine learning models at scale—to accelerate time to value—with the new HPE Machine Learning Development System. The new system, purpose-built for AI, is a complete solution that integrates a machine learning software platform, compute, accelerators, and network to develop and train more accurate AI models quickly and at scale.
Illustration: frame, presentation video The Big Shift: What is Swarm Learning? The Big Change: What is Distributed Learning?