HPE Swarm Learning

Industry’s first privacy preserving, decentralized machine learning solution.An%20analysis%20as%20of%20April%2013%2C%202022%20of%20competing%20offerings%20that%20claim%20privacy%20preservation%20found%20that%20they%20use%20a%20federated%20architecture%20reliant%20on%20a%20central%20server.


Protect privacy of data

Models are run at or near the distributed data sources and only insight and results are shared with collaborating machine learning peers. Data is not moved from the sources thus preserving data privacy by limiting data movement.

Improve accuracy and efficiency of models

Machine learning at the edge or near edge the data remains at the source preventing the inefficient movement of data or data duplication to the core or central location.

Ensure collaboration and enable better business decisions

Leverage the security of blockchain smart contracts to work collaboratively with peers and improve model insights.

Increase accuracy and reduce biases in AI models with HPE Swarm Learning

HPE Swarm Learning is a decentralized, privacy-preserving framework for performing machine learning model training at the data source. With HPE Swarm Learning, the data stays local and only the learnings are shared, resulting in improved models with less bias, while solving for data privacy, data ownership and efficiency concerns. Further, HPE Swarm Learning leverages a permissioned blockchain to securely onboard members, and dynamically elect the leader – providing resilience and security to the swarm network.

+ show more
Swarm Learning technology

Swarm Learning explained

Drawing its inspiration from biology, swarm learning is a decentralized machine learning solution built on blockchain technology, particularly designed to enable enterprises to harness the power of distributed data while protecting data privacy and security. Swarm learning leverages the computing power at or near the distributed data sources, ensures security using tested blockchain technology and protects privacy by sharing insights captured from machine learning and deep learning models running at the source data instead of the raw data itself.

Learn more about Swarm Learning technology

Swarm learning is the next gold rush for machine learning - training at the edge so the edge devices get smarter and also train their peers. With no central authority, blockchain is integrated to add control, privacy, and security.

Introducing HPE Swarm Learning

Decentralized privacy-preserving machine learning at the data source

HPE Swarm Learning provides efficient, secure, privacy-preserving, decentralized machine learning at the data source by sharing only insights among machine learning peers, not privileged data.

HPE Swarm Learning is decentralized, privacy-preserving machine learning framework. This framework utilizes the computing power at, or near, the distributed data sources to run the machine learning algorithms that train the models. It uses a blockchain platform to share only the learnings with peers collaboratively, improving insights. Training the model occurs at the edge, where data is most recent, and where prompt, data-driven decisions are necessary. The insights learned are shared with the collaborating peers, not the raw data thus preserving data privacy and limiting data movement, improving efficiency.

View the blog
Learn more about HPE Swarm Learning

Are data privacy and ownership constraints making your AI Models inaccurate and susceptible to bias? Are you concerned about data privacy, regulations, compliance, or moving data for your machine learning? Is your data distributed and preventing you from collaboration for improved and accurate models? Discover how HPE Swarm Learning can give your organization a competitive edge.

HPE Swarm Learning key benefits

HPE Swarm Learning for the competitive edge

HPE Swarm Learning  started  with  the  purpose  of  solving  the  dilemma  of  the  explosion of  data  and  the  technical,  social,  and  economic  challenges of  extracting  values  from  data that was constrained by privacy and ownership rules.  Organizations  embracing  HPE Swarm  Learning can improve the accuracy of their AI models while reducing biases in the same and enabling a more efficient AI infrastructure. Key benefits include:

  • Privacy and security
    By sharing only the ML insights, not actual data, between ML peers, HPE Swarm Learning  helps  businesses with privacy  regulations  by  giving  data  owners  greater  control  over  access  to and usage  of  their  data  through the  smart  contracts  of  blockchain,  as  well  as  eliminating  the  need  for  raw  data transfer over networks, boosting customer  confidence.
  • Efficiency
    HPE Swarm Learning enables the reduction of the costs  of  raw  data  transfer  by  performing  learning  at  or  near  the  data sources.  HPE Swarm Learning can further help reduce  operation  costs  by  taking  advantage  of  existing  storage  and  computing  capabilities  at  or  near  the  data sources,  thus  eliminating  the  investment  in a  central  facility, either  on-premises  or  in  a  cloud.
  • Fault-Tolerance for ML Models
    Compared to centralized model training, HPE  Swarm  Learning  decentralizes    learning,  thus  effectively  avoiding  a single  point  of  failure,  which can  threaten  business  continuity.  The  HPE Swarm  Learning  algorithm  is  effective  in  handling  biased  and  unbalanced data  at the  source. Block chain adds  robustness  in handling  exceptions  such  as  the  lost  connection  of  a  data  source and local model  to  its   machine learning  peers.
  • Timely Insights
     HPE Swarm Learning brings  with  it  the powerful  benefit  of  reducing  the latency between  the  creation  of  data  and the availability  of  actionable insight  derived  from  that  data.  With HPE  Swarm Learning,  model  retraining  can be  initiated  as  soon  as  new  data becomes  available  at  any  data source.  The learning captured  can be  shared  immediately  with all  the  machine learning  peers,  without  waiting  for  the  data  to  be  transferred, consolidated,  and  then mined.  A shorter path between  data and  insights  means  faster  and  more accurate responses  to  the ever-changing market,  an  enviable  competitive  advantage.
  • New collaboration models
    By drawing a  clear  line  between  raw  data  and  the insights  derived from  that  data,  HPE  Swarm  Learning  decouples  data  access  and  data ownership.  This decoupling, along  with  the  shifting  of  training  to data,  provides  a holistic  perspective  on  the  value  of  data. 
Learn more about HPE Swarm Learning

Find the right HPE Swarm Learning solution for your organization.

Key technical specifications

Decentralized privacy-preserving machine learning software

  • Gives data owners better control over access to their sensitive information
  • Extracts insight from the source data without invading data privacy
  • Facilitates the leveraging of any data from any data source worldwide and may help in complying with privacy regulations
  • Improves models by leveraging larger data set and generating a model with global stage merge
  • Web UI – interface for ease of installation

Blockchain-based smart contract between participants

  • Improved efficiency as data is not transferred to a central location
  • Prevents inside attack on model merging

Fully supported containerized software

  • Packaged as containers with 3 SKUs of 5, 10, 20 nodes of 1 year subscription
  • Each 5-pack container can be configure
    - 5 HPE Swarm Learning servers optionally GPU based
    - 1 Management Server (non-GPU based)
  • Community Edition- No charge, unlimited term,  5 node, best effort support available

Fault-tolerance for AI models

  • Once it is set up or has connectivity any node in the network that fails joins the network and continues participation from where it left off

Software container contents

  • Swarm Learning – for running user-defined machine learning algorithm
  • Swarm Network – the blockchain network
  • License server – Autopass license server
  • SWOP (Swarm Operator) – for command-line management
  • 1-yr subscription license and a community license that supports up to 5 nodes

Supported hardware

  • HPE Swarm Learning can run on any hardware that supports executing container software
    - HPE Edgeline
    - HPE ProLiant
    - HPE Superdome Flex
    - HPE Apollo
  • HPE Swarm Learning is qualified with DL 380/5, Apollo 6500
  • PFSS Storage if external storage is required

Professional Services

HPE Advisory and Professional Services define the Edge AI adoption roadmap including driving use cases and technology needs. Prove the value of edge AI adoption and scale its deployment across the swarm entities. Building a productive collaboration swarm requires data preparation, models porting and monitoring, HPE Pointnext experts take care of these critical tasks and let the Customer focus on model development and transforming insights into actions.

HPE PointNext 24x7 Tech Care Essential support

  • Break fix support
  • Backed by HPE

Explore HPE Artificial Intelligence topics

Take advantage of our comprehensive solutions to untangle complexity and create your end-to-end artificial intelligence solution, from the core data center to the intelligent edge.

Swarm Learning resources

Want to explore more? Here are some swarm learning resources that might interest you.