Personal Information Detection via Multi-Stage Filtering and Model Update
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Supervised learning models can produce incorrect predictions for input data that can be clearly distinguished, leading to inaccurate personal information detection.
Innovation Solution
A personal information detection reinforcement method using multiple filtering, which involves first filtering input data using record and pattern data, classifying it with a supervised learning model, performing second filtering using an unsupervised-based algorithm, and updating the model based on the filtered results to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If supervised learning model is used for personal information detection, then classification can be performed based on learned patterns, but wrong predictions occur even for clearly distinguishable input data
Solution Approach 1:
The detection system is segmented into multiple independent filtering stages: first filtering using record data and pattern data, then supervised learning classification, followed by second filtering using unsupervised algorithms. Each segment handles specific aspects of detection, preventing single-point failures and improving overall accuracy through distributed decision-making.
Solution Approach 2:
Two filtering mechanisms are introduced as intermediaries between raw input and final classification. The first filter preprocesses data using record and pattern matching, while the second filter post-processes results using unsupervised algorithms. These intermediaries mediate between the supervised learning model and input data, correcting errors and improving prediction accuracy.
2Measurement precision
If multiple filtering stages are applied to improve prediction accuracy, then wrong predictions are corrected, but processing time and system complexity increase
Solution Approach 1:
The first filtering stage performs preliminary action by preprocessing input data using record data and pattern data before the supervised learning model processes it. This preliminary filtering removes obvious cases and prepares data in advance, reducing the burden on subsequent stages and improving overall efficiency despite added complexity.
Solution Approach 2:
The system implements feedback mechanisms where the second filtering stage uses unsupervised algorithms to analyze the outputs of the supervised learning model and identify incorrect predictions. This feedback loop continuously corrects errors and improves prediction accuracy, justifying the increased system complexity through enhanced reliability.
3Measurement precision
If multiple filtering stages are applied to improve prediction accuracy, then wrong predictions are corrected, but processing time increases
Solution Approach 1:
The system applies partial filtering action by using the first filter to handle obvious cases using record and pattern data, allowing the supervised learning model to focus on more complex cases. This partial application of filtering reduces unnecessary processing time while maintaining high accuracy for clearly distinguishable inputs.
Solution Approach 2:
Different filtering stages apply different levels of processing quality to different types of input data. The first filter applies rapid pattern matching for simple cases, while the supervised learning model and second filter provide more thorough analysis for complex cases. This local quality adjustment optimizes processing time by avoiding excessive filtering for already-clear cases.
Data Source
AI summary
Disclosed are a personal information detection reinforcement method using multiple filtering and a personal information detection reinforcement apparatus using the same. The personal information detection reinforcement method includes performing first filtering of input data using record data and pattern data, classifying a class of the first-filtered input data using a previously constructed supervised learning model, performing second filtering of the first-filtered input data using an unsupervised-based algorithm based on the classified class, and updating the supervised learning model based on the second-filtered result data.