Machine Learning Detection of Miscoded Data-Field Tags
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection systems in financial journal entries are labor-intensive and require constant maintenance to adapt to changing business needs, failing to identify complex patterns and prone to human error.
Innovation Solution
A two-stage machine learning solution using a Bayesian network and Hyperband algorithm to identify anomalies, iteratively remove fields, and generate permutations to find optimal replacements, reducing human intervention and adapting to data changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If rules-based systems are used to detect anomalies in journal entries, then mechanical errors can be caught, but the systems require constant maintenance and cannot identify complex patterns
Solution Approach 1:
The patent replaces traditional rules-based mechanical systems with a machine learning model that automatically learns patterns from data. The ML model substitutes manual rule creation and maintenance with automated pattern recognition, enabling the system to identify both simple and complex anomalies without requiring constant human intervention to update rules.
Solution Approach 2:
The machine learning model performs self-learning and self-adjustment by continuously training on new data patterns. The system automatically adapts to changing business needs and data characteristics without requiring external reconfiguration, making the system self-maintaining and reducing the burden on professionals to update anomaly detection rules.
2Measurement precision
If traditional accounting and auditing processes are used to review journal entries, then thorough investigation can be performed, but the processes are labor-intensive and require small armies of professionals
Solution Approach 1:
The patent replaces manual accounting and auditing processes with an automated machine learning-based system. The ML model processes journal entries automatically, performing thorough anomaly detection and investigation without human intervention. This substitution maintains or improves detection quality while dramatically increasing processing capacity and eliminating the need for large teams of professionals.
Solution Approach 2:
The machine learning model performs multiple functions that previously required different professionals: anomaly detection, pattern recognition, investigation, and correction suggestion. This multi-functional system consolidates the capabilities of multiple specialists into a single automated platform, improving both thoroughness and efficiency simultaneously.
3Adaptability or versatility
If rules-based systems are deployed to catch misclassification errors, then initial mechanical errors can be identified, but the systems fall short in identifying non-trivial patterns and require constant intervention
Solution Approach 1:
The machine learning model continuously self-trains on new data, automatically adapting to emerging patterns and business changes. This self-service capability eliminates the need for constant human intervention to update detection rules, as the system autonomously learns and adjusts to new misclassification patterns and business requirements.
Solution Approach 2:
The system performs preliminary learning and pattern recognition by training on historical data before deployment. This preliminary action enables the model to proactively identify complex patterns and adapt to business needs in advance, reducing the frequency of subsequent interventions required to maintain effective anomaly detection.
Data Source
AI summary
The techniques described herein relate to a method including: receiving, by a processor, a data record having a plurality of fields; generating, by the processor, a risk score for the data record using a predictive model; determining, by the processor, that the data record is a potential anomaly based on the risk score; identifying, by the processor, an anomalous field from the plurality of fields; generating, by the processor, a plurality of permutations of the data record, the plurality of permutations generated by changing a value of the anomalous field; and outputting, by the processor, a replacement record selected from the plurality of permutations, the replacement record having a field value for the anomalous field that generates a lowest risk score among the plurality of permutations.