Text-Based Recognition Model Training for Simulated Behavior Accuracy

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing simulated user behavior information generated through artificial intelligence programs is often inaccurate, leading to unrealistic user behavior patterns that hinder effective customization and development of intelligent device applications.

Innovation Solution

A model training method that converts behavior sequence data into text data, rearranges attribute data order, and trains a recognition model to minimize deviation from actual behavior data, determining the authenticity of simulated behavior information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If simulated behavior information is generated through artificial intelligence programs, then development efficiency is improved, but accuracy of behavior information deteriorates

Engineering Contradiction:

Improvedevelopment efficiencyVSAvoidaccuracy of behavior information

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent introduces a feedback mechanism where the recognition model evaluates simulated behavior information against actual user behavior patterns. The model continuously refines its recognition accuracy through training on real behavior data, enabling developers to identify and correct inaccuracies in simulated behavior information while maintaining efficient development workflows.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent replaces traditional simulation methods with a recognition model-based verification system. Instead of relying solely on AI-generated simulations, the system uses a trained recognition model to authenticate behavior sequences by comparing them against actual user behavior patterns, thereby improving accuracy without sacrificing development efficiency.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Loss of time

If simulated behavior information is used for application development, then development time is reduced, but reliability of behavior data deteriorates

Engineering Contradiction:

Improvedevelopment timeVSAvoidreliability of behavior data

Core Design Contradiction:

Loss of timeVSReliability

Solution Approach 1:

The patent applies preliminary action by pre-training the recognition model on actual user behavior data before it is used to verify simulated behavior information. This preliminary training ensures that the model has learned accurate behavior patterns from real users, enabling it to reliably distinguish between authentic and simulated behavior sequences during the development process.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The recognition model provides feedback to developers about the reliability of simulated behavior information. By comparing simulated behavior sequences against patterns learned from actual user data, the model identifies unreliable or inaccurate simulations, allowing developers to focus only on verified reliable data while maintaining reduced development time.

Inventive Principle:

Principle #23Feedback

3Measurement precision

If behavior sequence data is converted into text data for model training, then model training accuracy is improved, but data processing complexity increases

Engineering Contradiction:

Improvemodel training accuracyVSAvoiddata processing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces text data as an intermediary representation between raw behavior sequence data and the recognition model. By converting behavior sequences into text format, the system creates a bridge that enables the model to process and analyze behavior patterns more effectively, improving training accuracy while managing processing complexity through structured text representation.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20250245451A1Model training methods and apparatuses and task execution methods and apparatuses

Publication Date: 2025.07.31 ALIPAY (HANGZHOU) INFORMATION TECH CO LTD

AI summary

This specification discloses methods and apparatuses for model training and a task execution. An example model training method includes: obtaining behavior sequence data; converting event data of each behavior event included in the obtained behavior sequence data into text data, to obtain behavior text data corresponding to the behavior sequence data; then, inputting the obtained behavior text data into a to-be-trained recognition model, so that the to-be-trained recognition model outputs a recognition result for the behavior sequence data as a to-be-verified result based on the input behavior text data; and training the to-be-trained recognition model with an optimization objective of minimizing a deviation between the to-be-verified result output by the recognition model and an actual recognition result corresponding to the behavior sequence data.