Neural Network Data Identification for Sensitive Pattern Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for identifying sensitive data in datasets are inefficient and inaccurate, particularly in recognizing un-recognized patterns and characteristics, due to reliance on predefined rules and regular expressions that may not capture all patterns, leading to incomplete or incorrect identification.
Innovation Solution
A system utilizing deep neural networks, specifically Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), with adaptive pooling layers, to process and identify Personally Identifiable Information (PII) by converting input data into suitable formats for pattern recognition, enabling differentiation between various types of PII and non-PII attributes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If regular expressions-based matching and look up against reference master list are used to identify sensitive data, then the identification process is simple and fast, but the accuracy is low and unrecognized patterns cannot be detected
Solution Approach 1:
The patent replaces the mechanical rule-based system (regular expressions and look-up tables) with a neural network-based intelligent system. The neural network learns patterns automatically from training data, enabling detection of unrecognized patterns and complex relationships that predefined rules cannot capture, thereby improving identification accuracy while maintaining system feasibility through automated learning.
Solution Approach 2:
The patent transforms the identification approach from static parameter matching (predefined patterns and rules) to dynamic parameter learning (neural network weights and biases adjusted during training). This allows the system to adapt to different data characteristics and improve accuracy by learning optimal parameters from training data rather than relying on fixed rules.
2Adaptability or versatility
If predefined rules are used to identify patterns in data, then the identification process is straightforward, but unrecognized patterns and characteristics remain un-identifiable
Solution Approach 1:
The patent performs preliminary training of the neural network model using historical data before actual sensitive data identification. This preliminary action allows the system to learn patterns and characteristics in advance, making it adaptable to various data types and patterns. The training phase, though time-consuming, is performed once or periodically, enabling fast and accurate identification during operational use.
3Reliability
If existing techniques are used to tag sensitive data, then the process is simple, but false positives and negatives occur due to similar patterns in different categories
Solution Approach 1:
The patent segments the identification task into multiple neural network layers that process different features and aspects of the data independently. Each layer learns specific patterns and characteristics, allowing the system to distinguish between similar patterns in different categories by analyzing multiple segmented features rather than relying on a single simple rule.
Data Source
AI summary
Examples of determining a classification for an input dataset are provided. The input dataset may be defined in a one-dimensional data structure. The input data set may be converted into a formatted dataset of a two-dimensional data structure, a format of the formatted dataset being defined in accordance to a type of a deep neural network component. The formatted dataset may be processed through multiple layers of the deep neural network component. Based on the processing of the formatted dataset, a classification indicative of a probability of a data feature of the input dataset corresponding to an identity parameter, which may include sensitive data, associated with an identity of the individual, may be determined. A user may be provided the data feature of the input dataset corresponding to the identity parameter in a first format and another data features of the input dataset in a second format different than the first format.