Polynomial Curve Coefficients for Fraud Detection Model Data Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fraud detection models using machine learning face challenges in efficiently monitoring and comparing performance data across different time periods due to large data sets and misalignment of independent variable values, leading to network bandwidth burdens and inefficient optimization processes.
Innovation Solution
A computerized method employing data compression using polynomial regression to generate coefficients, which are then adjusted with a machine learning model to optimize fraud detection model performance, allowing for efficient transmission and comparison of compressed data across a computer network, enabling automated assessment and reduced computational power.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If performance data is transmitted across computer network for monitoring, then fraud detection model performance can be monitored and optimized, but network bandwidth is consumed and data transmission becomes burdensome
Solution Approach 1:
The patent extracts only the essential information from performance data by fitting a polynomial curve through the data points and transmitting only the curve coefficients rather than the complete raw data. This extraction maintains the core performance monitoring capability while dramatically reducing the amount of data that needs to be transmitted across the network.
Solution Approach 2:
The patent changes the representation of performance data from raw data points to polynomial curve coefficients. By transforming the data into a different parameter space (from (x,y) data points to polynomial coefficients), the system achieves data compression that reduces network bandwidth requirements while preserving the ability to monitor and compare performance trends.
2Productivity
If performance data with many data points is transmitted for analysis, then performance trends can be identified, but data transmission becomes burdensome and processing complexity increases
Solution Approach 1:
The patent extracts the trend information from the raw performance data by fitting a polynomial curve that captures the overall pattern. The curve coefficients serve as a compressed representation that retains the essential trend information while eliminating the need to process and transmit the complete raw data set.
Solution Approach 2:
The patent segments the performance data into discrete polynomial coefficients that can be independently transmitted and processed. This segmentation allows the complex performance data to be broken down into manageable components that are easier to transmit and analyze, reducing overall processing complexity.
3Measurement precision
If independent variable values are aligned exactly across time periods for comparison, then accurate performance comparison is achieved, but data processing complexity and time requirements increase
Solution Approach 1:
The patent changes the approach from aligning data points to comparing polynomial curves. By transforming the data into curve representations, the system can compare performance across time periods without requiring exact alignment of independent variable values, as the polynomial coefficients capture the overall trend patterns that remain comparable even when timing differs.
Data Source
AI summary
Fraud detection model performance is represented as compressed data in the form of polynomial curve coefficients. A data compression setting, a set of independent variable values, and a set of dependent variable values are used in polynomial regression to generate coefficients of a polynomial curve. The data compression setting is related to the order of the polynomial, for example set to the degrees of freedom (DOF) defined as the polynomial order. A lower DOF yields a higher error, but with a higher degree of compression. The lowest DOF with a tolerable error is selected and the polynomial coefficients are transmitted to a remote node. The remote node regenerates the polynomial curve for comparison with a polynomial curve from a prior time period, in order to determine a performance trend. The trend is used to either generate an alert or trigger further training of the fraud detection model.