Machine Learning Model for Trait-Intersection Count Estimation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional data-management systems face inefficiencies and inaccuracies in computing digital-intersection counts from large data repositories, leading to excessive resource consumption and delayed reporting, with limitations in applying these counts to other metrics.

Innovation Solution

A machine-learning model is employed to analyze semantic-trait embeddings and initial trait-intersection counts to efficiently and accurately estimate trait-intersection counts for a target time period, reducing computational burden and improving accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional data-management systems count client-device users one-by-one from a database, then they can provide accurate trait-intersection counts, but they consume excessive processor and computing resources and experience significant delays

Engineering Contradiction:

Improvetrait-intersection count accuracyVSAvoidcomputing efficiency

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent replaces the mechanical one-by-one counting process with a machine-learning model that performs estimation. Instead of iteratively processing each user record from the database, the system uses trained ML models to predict trait-intersection counts based on input traits and time periods, dramatically improving computing efficiency while maintaining acceptable accuracy

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the approach from exact counting to estimated counting using machine learning. The system trains ML models on historical data and uses them to predict trait-intersection counts, transforming the problem from a deterministic counting task to a probabilistic estimation task that is computationally more efficient

Inventive Principle:

Principle #35Parameter changes

2Productivity

If conventional data-management systems rely on sample extrapolation to preserve computing resources, then they reduce resource consumption, but they introduce significant counting errors

Engineering Contradiction:

Improvecomputing efficiencyVSAvoidtrait-intersection count accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent replaces the statistical extrapolation method with a machine-learning-based estimation system. Instead of using simple sample-based projections, the system employs trained ML models that learn complex patterns from historical data, providing more accurate estimates without requiring full database processing

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent performs preliminary training of machine-learning models on historical trait-intersection data before deployment. This pre-computation phase allows the models to learn patterns and relationships in advance, enabling accurate real-time or near-real-time estimation without processing the entire database during query execution

Inventive Principle:

Principle #10Preliminary action

3Adaptability or versatility

If conventional data-management systems count users sharing characteristics exclusively for reporting, then they provide basic count reports, but they cannot extend or apply the counts to other metrics

Engineering Contradiction:

Improveapplication extensibilityVSAvoidsystem complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent creates a universal machine-learning model that can estimate trait-intersection counts for any combination of traits and time periods. The same model infrastructure serves multiple functions: generating count estimates, supporting different traits, handling various time periods, and enabling downstream applications beyond basic reporting

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent introduces a machine-learning model as an intermediary layer between the raw database and the counting function. This intermediary transforms the system from a simple counting tool to a versatile estimation platform that can handle complex queries and support multiple applications without directly accessing or processing the entire database

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS11429653B2Generating estimated trait-intersection counts utilizing semantic-trait embeddings and machine learning

Publication Date: 2022.08.30 ADOBE INC

AI summary

This disclosure relates to methods, non-transitory computer readable media, and systems that, upon request for a trait-intersection count of users (or other digital entities) corresponding to traits for a target time period, use a machine-learning model to analyze a semantic-trait embedding of the traits and to generate an estimated trait-intersection count of such entities sharing the traits for the target time period. By applying a machine-learning model trained to estimate trait-intersection counts, the disclosed methods, non-transitory computer readable media, and systems can analyze both a semantic-trait embedding of traits and an initial trait-intersection count of trait-sharing entities for an initial time period to estimate the trait-intersection count for the target time period. The disclosed machine-learning model can thus analyze both the semantic-trait embedding and the initial trait-intersection count to efficiently and accurately estimate a trait-intersection count corresponding to a requested time period.