Human-Machine Identification via Behavioral Characteristic Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for distinguishing human and machine interactions, such as CAPTCHA tests, are not sufficiently accurate in differentiating between human and machine behaviors, particularly in recognizing complex patterns and interactions.
Innovation Solution
A behavioral characteristic model is trained using machine learning, which extracts and filters operation samples based on movement speed and randomness, reduces dimensions using PCA, and classifies data using SVM to differentiate between human and machine behaviors through specific behavioral characteristics like operative slide time length, center line-deviation frequency, and average deviation distance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional CAPTCHA tests are used to distinguish human and machine interactions, then the system can implement basic security verification, but the accuracy in differentiating between human and machine behaviors is insufficient
Solution Approach 1:
The system performs preliminary filtering of operation samples based on movement speed and randomness characteristics before main classification. This pre-processing step eliminates obviously machine-generated operations early, improving overall identification accuracy while maintaining reasonable system complexity
Solution Approach 2:
The patent applies Principal Component Analysis (PCA) to reduce the dimensionality of behavioral feature data while preserving essential discrimination information. This transforms high-dimensional operation data into a lower-dimensional space that maintains identification accuracy but reduces computational complexity
2Measurement precision
If complex behavioral analysis is performed to improve identification accuracy, then human-machine differentiation improves, but processing time increases
Solution Approach 1:
The identification process is divided into multiple stages: initial filtering based on simple criteria (movement speed, randomness), followed by dimensionality reduction, and finally classification. This segmentation allows quick rejection of obvious machine operations while applying more complex analysis only when necessary, reducing average processing time
Solution Approach 2:
The system transforms raw operational parameters into derived behavioral features (e.g., movement speed, randomness, operative slide time length, center line-deviation frequency). This parameter transformation enables more efficient classification by converting complex interaction patterns into quantifiable metrics that can be processed faster
Data Source
Figure 1A
Figure 1B
Figure 1C
AI summary
Identifying a human or machine is disclosed including sending a currently-triggered human or machine identification request to a server, receiving a first operation rendered by the server in response to the human or machine identification request, the first operation being configured to identify a human or machine and expressing an operation track for identifying the human or machine, and sending a second operation that was triggered in response to the first operation to the server to identify the human or machine on a triggering entity for the second operation based on a behavior characteristic model that was set in advance.