Swarm Learning Parameter Merging With Homomorphic Encryption

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Managing geographically distributed, decentralized enterprise systems with heterogeneous nodes that operate outside an enterprise network poses challenges due to security concerns, data privacy, and reliance on a central controller node, which creates a single point of failure and scalability issues.

Innovation Solution

Implementing a swarm learning platform using homomorphic encryption and blockchain technology to elect a merge leader for decentralized parameter merging, ensuring secure data sharing without revealing raw data, and utilizing a decentralized management system without a central controller, enabling fault-tolerant and scalable model training across multiple nodes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If a central controller node is used to manage decentralized systems, then system coordination is simplified, but reliability deteriorates due to single point of failure

Engineering Contradiction:
Improvesystem coordinationVSAvoidfault tolerance
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The system segments the central controller function into multiple decentralized nodes. Each node independently manages its own operations and participates in consensus protocols, eliminating the single point of failure while maintaining coordinated system operation through distributed decision-making mechanisms.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces blockchain technology as an intermediary layer that enables trust and coordination between decentralized nodes without requiring a central controller. The blockchain serves as a shared ledger and consensus mechanism that mediates interactions between nodes, ensuring reliable operation through cryptographic proof and distributed validation.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If raw data is shared for model training, then communication efficiency is improved, but data privacy deteriorates

Engineering Contradiction:
Improvecommunication efficiencyVSAvoiddata privacy risk
Core Design Contradiction:
ProductivityVSObject-affected harmful factors

Solution Approach 1:

The patent employs homomorphic encryption as an intermediary that enables computations on encrypted data without decrypting it. The encryption scheme allows model parameters to be processed and merged while remaining encrypted, with only the final result being decrypted. This maintains communication efficiency for parameter sharing while preserving data privacy by ensuring raw data never leaves its local node in plaintext form.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system replaces the mechanical approach of sharing raw data for training with a cryptographic substitution using homomorphic encryption. Instead of transmitting and processing plaintext model parameters, the system transmits and processes encrypted parameters, substituting the traditional data sharing mechanism with a secure cryptographic one that maintains computational functionality while eliminating privacy risks.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Reliability

If decentralized parameter merging is implemented, then reliability is improved, but device complexity increases

Engineering Contradiction:
Improvefault toleranceVSAvoidsystem architecture
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements universal cryptographic primitives and standardized protocols that enable multiple nodes to perform consistent encryption, computation, and verification functions. By using standardized homomorphic encryption schemes and consensus algorithms, the system achieves decentralized parameter merging without requiring each node to implement custom complex logic, thereby reducing overall system complexity while maintaining fault tolerance.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The blockchain serves as a mediating infrastructure that handles the complexity of coordination, consensus, and trust management between nodes. By offloading these complex coordination tasks to the blockchain layer, individual nodes don't need to implement complex inter-node communication and verification logic, reducing the complexity burden on each individual device while maintaining reliable decentralized operation.

Inventive Principle:
Principle #24Intermediary (Mediator)

4Adaptability or versatility

If geographically distributed nodes are used, then adaptability is improved, but management difficulty increases

Engineering Contradiction:
Improvegeographic distributionVSAvoidmanagement complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent uses blockchain as an intermediary management layer that provides unified coordination for geographically distributed nodes. The distributed ledger and consensus mechanisms enable nodes across different locations to maintain consistent state and execute coordinated operations without requiring complex centralized management infrastructure. This intermediary layer abstracts the geographic distribution complexity, allowing adaptable node placement while simplifying management through standardized cryptographic protocols.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3935776B1Secure parameter merging using homomorphic encryption for swarm learning
Publication Date: 2025.08.06 HEWLETT PACKARD ENTERPRISE DEV LP
  • EP3935776B1 patent drawingFigure 1A
  • EP3935776B1 patent drawingFigure 1B~1C
  • EP3935776B1 patent drawingFigure 1D

AI summary

Systems and methods are provided for implementing swarm learning while using blockchain technology and election/voting mechanisms to ensure data privacy. Nodes may train local instances of a machine learning model using local data, from which parameters are derived or extracted. Those parameters may be encrypted and persisted until a merge leader is elected that can merge the parameters using a public key generated by an external key manager. A decryptor that is not the merge leader can be elected to decrypt the merged parameter using a corresponding private key, and the decrypted merged parameter can then be shared amongst the nodes, and applied to their local models. This process can be repeated until a desired level of learning has been achieved. The public and private keys are never revealed to the same node, and may be permanently discarded after use to further ensure privacy.