Unified ML and Rules Platform via Directed Acyclic Graph
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current software applications face challenges in auditing and efficiently developing complex logic, as they often require separate processes and resources for machine learning models and rules, making it difficult to identify and adjust components that produce incorrect results.
Innovation Solution
Integration of machine learning models and rules within a unified platform using a directed acyclic graph (DAG) that allows for iterative feedback loops to modify and improve both models and rules, enhancing auditability and efficiency by sharing resources and logic paths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If machine learning models and rules are developed and executed as independent components, then each component can be specialized and optimized independently, but the system complexity increases and auditability decreases due to separate development processes and resources
Solution Approach 1:
The patent combines machine learning models and rules into a unified platform where both components share common infrastructure, data access, and execution environments. This integration reduces system complexity while maintaining the specialized functions of each component, thereby improving auditability without sacrificing optimization capabilities.
Solution Approach 2:
The unified platform provides universal services to both machine learning models and rules, including data preprocessing, feature engineering, and execution management. This multi-functional approach eliminates redundant components and simplifies the overall system architecture while enhancing traceability and auditability across different logic types.
2Ease of manufacture
If complex logic is implemented using separate machine learning and rules components, then each component can be optimized independently, but it becomes difficult to determine which aspects of the logic should be adjusted when incorrect results are produced
Solution Approach 1:
The unified platform implements comprehensive feedback mechanisms that track the execution flow, data transformations, and decision outcomes across both machine learning models and rules. When incorrect results occur, the system can trace back through the integrated logic paths to identify whether the issue originates from model predictions, rule evaluations, or their interactions, significantly easing root cause analysis.
Solution Approach 2:
The platform introduces intermediary components that bridge machine learning models and rules, including shared feature stores, unified decision engines, and integrated monitoring systems. These intermediaries facilitate coordinated adjustment of logic components by providing common interfaces and standardized data formats, making it easier to modify and test individual components within the broader system context.
3Productivity
If separate processes and resources are used for developing machine learning models and rules, then each component can be developed independently, but the overall development efficiency decreases due to duplicated efforts and resources
Solution Approach 1:
The patent merges the development environments, toolchains, and resource pools for machine learning models and rules into a unified platform. This consolidation eliminates duplicated infrastructure, shared common libraries and frameworks, and streamlined workflows that allow developers to work on both component types within the same ecosystem, thereby improving development efficiency while reducing overall resource consumption.
Data Source
AI summary
Aspects of the present disclosure provide techniques for machine learning and rules integration. Embodiments include receiving input values corresponding to a subset of a set of input variables associated with an automated determination. Embodiments include generating a directed acyclic graph (DAG) representing a set of constraints corresponding to the set of input variables. The set of constraints relate to one or more machine learning models and one or more rules. Embodiments include receiving one or more outputs from the one or more machine learning models based on one or more of the input values. Embodiments include determining outcomes for the one or more rules based on at least one of the input values. Embodiments include populating the DAG based on the input values, the one or more outputs, and the outcomes. Embodiments include making the automated determination based on logic represented by the DAG.