Micro-expression Detection for Fraud Prevention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques fail to effectively identify the mental state of authenticated users and prevent fraudulent actions, as they struggle to distinguish between genuine and fraudulent intentions or states, such as intoxication, using visual micro-expressions.
Innovation Solution
A detection device that acquires imaging videos of users, specifies categories of micro-expressions, and determines whether their appearance patterns are normal by learning models from large datasets to predict and compare against predetermined patterns, thereby identifying abnormal mental states and preventing fraudulent access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional techniques use visual micro-expression monitoring to detect deepfakes, then they can distinguish real persons from still images, but they fail to identify mental states and prevent fraudulent acts by authenticated users
Solution Approach 1:
The patent segments the detection process into multiple independent modules: micro-expression detection module, mental state identification module, and fraudulent act prevention module. Each module handles a specific aspect of the detection task, allowing the system to maintain high detection accuracy while expanding functional scope to cover both deepfake detection and mental state analysis.
Solution Approach 2:
The system is designed with multi-functionality to perform both deepfake detection and mental state identification using the same micro-expression monitoring technology. By making the detection system universal, it can authenticate users and prevent fraudulent acts beyond just distinguishing real persons from still images, thereby resolving the contradiction between detection accuracy and functional scope.
2Measurement precision
If the system monitors visually unrecognizable micro-expressions, then it can detect deepfakes, but it cannot identify mental states such as malicious intention or intoxication
Solution Approach 1:
The patent introduces an intermediary processing layer that analyzes micro-expression patterns as intermediate data to infer mental states. This intermediary analysis layer processes the detected micro-expressions and translates them into mental state information, enabling the system to recover lost mental state information while maintaining the original detection accuracy for deepfakes.
Solution Approach 2:
The system replaces traditional mechanical authentication methods with a sophisticated analysis system that uses machine learning algorithms to interpret micro-expression data. This substitution enables the system to extract mental state information from visual micro-expressions that were previously undetectable, thereby recovering lost information without compromising detection accuracy.
3Reliability
If the system analyzes micro-expression appearance patterns to determine normality, then it can identify abnormal mental states, but it requires complex processing of large datasets
Solution Approach 1:
The patent implements preliminary action by pre-processing and organizing micro-expression data into standardized appearance patterns before actual authentication. The system pre-establishes baseline patterns of normal micro-expressions and prepares analysis models in advance, which reduces processing complexity during real-time authentication while maintaining high reliability in identifying abnormal mental states.
Solution Approach 2:
The system applies parameter changes by transforming complex micro-expression data into simplified appearance pattern parameters that are easier to process. By changing the representation parameters of micro-expressions into standardized categories and features, the system reduces processing complexity while preserving the reliability needed to distinguish normal from abnormal mental states.
Data Source
AI summary
A detection device includes a processing circuit that is configured to acquire an imaging video of a user to be authenticated, specify categories of micro-expressions of the user using the acquired imaging video, and determine whether an appearance pattern of each specified category for the micro-expressions is normal.