Cross-Node Audience Propensity Learning in Privacy Clean Rooms

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing machine learning-based propensity prediction systems face challenges in implementing privacy and data security requirements while ensuring consistency, reliability, and accuracy of audience targeting for digital content distribution, particularly in scenarios where third-party data cannot be shared freely.

Innovation Solution

A protected execution environment with separate clean rooms for third-party nodes and an aggregator node is created, allowing third-party tuning of a first-party model without exposing data, and aggregating learnings to generate a globally-tuned version for further tuning and scoring, while maintaining data privacy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If third-party data is shared freely for model tuning, then model accuracy and reliability are improved, but data privacy and security are compromised

Engineering Contradiction:

Improvemodel accuracyVSAvoiddata privacy risk

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent introduces a trusted execution environment (TEE) as an intermediary between third-party data sources and the machine learning model. The TEE acts as a secure mediator that enables model tuning with third-party data while preventing unauthorized access to sensitive information. Data is processed within the encrypted TEE boundaries, allowing accuracy improvement without direct data exposure to external parties.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent creates a protected execution environment that functions as an inert or secure atmosphere for data processing. Within this TEE, data is handled in a controlled, isolated state that prevents harmful external influences and unauthorized access. The model tuning process occurs in this secure environment where data privacy is maintained while still enabling accurate model training.

Inventive Principle:

Principle #39Inert atmosphere (Inert environment)

2Adaptability or versatility

If third-party data is combined for aggregation, then model generalization is improved, but data security restrictions are violated

Engineering Contradiction:

Improvemodel generalizationVSAvoiddata security restriction

Core Design Contradiction:

Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

The TEE serves as an intermediary that enables safe aggregation of third-party data for model generalization. Multiple third-party data sources can contribute to model tuning within the TEE boundaries without direct data sharing between external parties. The secure environment mediates the aggregation process, allowing improved model adaptability while respecting data security restrictions.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent segments the data processing architecture into isolated third-party nodes, each with their own clean room environment within the TEE. This segmentation allows each party to contribute data independently without exposing their raw data to others. The model generalization benefit is achieved through aggregated learnings from segmented sources rather than direct data combination.

Inventive Principle:

Principle #1Segmentation

3Object-affected harmful factors

If a protected execution environment with separate clean rooms is implemented, then data privacy is maintained, but system complexity increases

Engineering Contradiction:

Improvedata privacy protectionVSAvoidsystem architecture complexity

Core Design Contradiction:

Object-affected harmful factorsVSDevice complexity

Solution Approach 1:

The patent implements a universal TEE architecture that can serve multiple third-party nodes simultaneously. Rather than creating separate isolated systems for each party, a single TEE infrastructure provides privacy protection for all participants. This multi-functional approach reduces overall system complexity compared to implementing individual secure environments for each third party.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent merges multiple third-party clean rooms into a unified TEE environment where privacy is maintained through logical isolation rather than physical separation. This consolidation reduces infrastructure complexity while preserving data privacy protections. The aggregated learnings from multiple sources are combined within the shared secure environment, simplifying the overall system architecture.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS12579467B2Decentralized cross-node learning for audience propensity prediction

Publication Date: 2026.03.17 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

Embodiments of the disclosed technologies receive a first-party trained model and a first-party data set from a first-party system into a protected environment, receive a first third-party data set into the protected environment, and, in a data clean room, joining the first-party data set and the first third-party data set to create a joint data set for the particular segment, tuning a first-party trained model with the joint data set to create a third-party tuned model, sending model parameter data learned in the data clean room as a result of the tuning to an aggregator node, receiving a globally tuned version of the first-party trained model from the aggregator node, applying the globally tuned version of the first-party trained model to a second third-party data set to produce a scored third-party data set, and providing the scored third-party data set to a content distribution service of the first-party system.