Federated Learning Partner Selection Using Client Similarity

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing federated learning systems struggle to select appropriate partners for machine learning models when clients are working on different tasks, leading to uncertainty about model accuracy.

Innovation Solution

An information processing apparatus and method that evaluates client-related information to determine the degree of similarity among clients, allowing for the selection of federation partners based on this similarity, thereby enhancing model accuracy in federated learning.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If federated learning is performed without evaluating client similarity, then the system can operate with any clients, but the model accuracy becomes uncertain when clients work on different tasks

Engineering Contradiction:

Improveability to work with different clientsVSAvoidmodel accuracy

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system performs preliminary evaluation of client-related information and calculates similarity scores before initiating federated learning. This advance assessment allows the system to identify suitable partners based on task compatibility, ensuring reliable model accuracy from the start rather than attempting corrections during the learning process.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary evaluation mechanism that assesses client similarity through calculated scores based on client-related information. This intermediary layer acts as a mediator between diverse clients and the federated learning process, filtering and matching clients appropriately before they participate together, thus maintaining both versatility and reliability.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If client-related information is collected and evaluated to determine similarity, then appropriate partners can be selected for federated learning, but the system complexity increases

Engineering Contradiction:

Improvemodel accuracyVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system extracts only the necessary client-related information required for similarity evaluation, separating these essential features from unnecessary data. By taking out only the relevant attributes needed for calculating similarity scores, the system maintains high model accuracy while avoiding the complexity burden of processing excessive information.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system transforms client-related information into standardized similarity score parameters that facilitate straightforward comparison and decision-making. By changing the parameters into a unified scoring system, the complex task of evaluating client compatibility is simplified into a clear numerical assessment, reducing system complexity while preserving reliability.

Inventive Principle:

Principle #35Parameter changes

3Ease of operation

If the system displays a selection screen with similarity information, then users can make informed decisions about partner selection, but the time required for partner selection increases

Engineering Contradiction:

Improveinformed decision-makingVSAvoidpartner selection time

Core Design Contradiction:

Ease of operationVSLoss of time

Solution Approach 1:

The system automatically calculates and presents similarity scores and partner recommendations, enabling users to make informed decisions without requiring them to manually analyze complex client data. The self-service approach provides ready-to-use information that guides user selection, maintaining ease of operation while minimizing the time users need to spend on the selection process.

Inventive Principle:

Principle #25Self-service

4Productivity

If federated learning is conducted without partner selection based on similarity, then the process is faster and simpler, but the resulting model accuracy deteriorates when clients have different tasks

Engineering Contradiction:

Improvefederated learning speedVSAvoidmodel accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system performs preliminary similarity evaluation and partner matching before the federated learning process begins. This advance preparation ensures that only compatible clients are grouped together, guaranteeing model accuracy from the start. The preliminary action eliminates the need for repeated learning cycles or corrections, thereby maintaining high productivity while ensuring reliability.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20260046220A1Information processing apparatus, information processing method, and non-transitory recording medium

Publication Date: 2026.02.12 RICOH CO LTD

AI summary

An information processing apparatus to communicate, via a network, with another information processing apparatus participating in federated learning includes circuitry to acquire client-related information generated based on information registered by a client from the another information processing apparatus, evaluate a degree of similarity between the client-related information on the another information processing apparatus and the client-related information on one or more partner candidates for the federated learning, and output screen data for displaying a screen allowing selection of one or more partners for the federated learning, based on the degree of similarity.