Sequence-Invariant Model for Low-Latency User Action Prediction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional machine learning systems introduce latency between event detection and user action influence, hindering accuracy and quality of outcomes, particularly in network-based advertisements and promotions.

Innovation Solution

A network computing system employs a sequence invariant model to analyze real-time user activities and events, reducing latency by predicting user intentions and facilitating immediate, user-specific interventions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If conventional machine learning systems are used to analyze user data, then processing can be performed, but latency is introduced between event detection and user action influence

Engineering Contradiction:

ImprovelatencyVSAvoiduser action influence speed

Core Design Contradiction:

Loss of timeVSProductivity

Solution Approach 1:

The system performs preliminary actions by pre-processing and storing user event data in a database before actual prediction is needed. The sequence invariant model is pre-trained on historical data, allowing rapid inference during real-time user interactions without performing heavy computation at the moment of prediction.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

A database serves as an intermediary between raw user event data and the prediction model. The database stores processed and formatted user data, allowing the model to quickly query relevant information without directly processing raw data streams, thus reducing latency between data collection and action prediction.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If real-time analysis is performed to reduce latency, then user engagement accuracy improves, but system complexity increases

Engineering Contradiction:

Improveuser action prediction accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments the user interaction process into distinct phases: data collection, data storage in database, model inference, and action generation. This segmentation allows each component to be optimized independently and simplifies the overall architecture by separating concerns between real-time processing and background data management.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Instead of processing all raw user data in real-time, the system creates a simplified representation (copy) of user behavior patterns stored in the database. The sequence invariant model then operates on this pre-processed copy, reducing computational complexity while maintaining prediction accuracy.

Inventive Principle:

Principle #26Copying

3Productivity

If sequence invariant model is used to predict user actions, then real-time engagement is enabled, but data processing requirements increase

Engineering Contradiction:

Improvereal-time user engagement efficiencyVSAvoiddata processing volume

Core Design Contradiction:

ProductivityVSQuantity of substance

Solution Approach 1:

The system extracts only the essential features from raw user event data and stores them in the database as simplified representations. The sequence invariant model then processes only these extracted features rather than the complete raw data streams, significantly reducing data processing requirements while maintaining prediction effectiveness.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS20250238696A1Network computer system using sequence invariant model to predict user actions

Publication Date: 2025.07.24 SESSION AI INC

AI summary

Embodiments provide for a computer system and method to employ a sequence invariant model to determine user intentions, based on monitoring of real-time activities of the user.