ML-Guided RPA Workflow Activation Using Confidence Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current RPA workflows are deterministic and may not be optimal for situations that change over time, lacking flexibility and adaptability.
Innovation Solution
Implementing machine learning (ML) models to enable probabilistic activities in RPA workflows, where confidence values determine whether to activate or deactivate workflow sections, allowing for hybrid deterministic and probabilistic logic, and enabling learning and improvement over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If deterministic logic is used in RPA workflows, then workflow execution is predictable and straightforward, but the system cannot adapt to changing conditions over time
Solution Approach 1:
The patent applies dynamics by transitioning from static deterministic workflows to dynamic probabilistic workflows that can adapt to changing conditions. ML models are integrated into workflow nodes to enable probabilistic branching based on predicted outcomes, allowing the workflow to dynamically adjust its execution path based on learned patterns and changing data characteristics.
Solution Approach 2:
The patent changes parameters by introducing probability thresholds and confidence levels into the workflow execution. Instead of fixed deterministic paths, the system uses probabilistic parameters from ML models to determine workflow progression. The patent also dynamically adjusts parameters such as confidence thresholds and probability cutoffs based on performance feedback and changing conditions.
2Measurement precision
If ML models are integrated into RPA workflows, then adaptability and accuracy improve, but system complexity and computational requirements increase
Solution Approach 1:
The patent segments the workflow into distinct deterministic and probabilistic sections. ML models are selectively applied to specific nodes where probabilistic decision-making is beneficial, rather than throughout the entire workflow. This segmentation allows the system to maintain simplicity in deterministic portions while adding intelligence only where needed.
Solution Approach 2:
The patent introduces an intermediary layer between deterministic workflow logic and ML models. This intermediary manages the integration by handling model inference, interpreting probability outputs, and translating them into workflow decisions. It abstracts the complexity of ML operations from the core workflow logic, making the system more manageable.
3Adaptability or versatility
If probabilistic activities are implemented, then workflow flexibility increases, but execution time and computational resources increase
Solution Approach 1:
The patent applies partial action by implementing probabilistic logic only in specific workflow nodes where it provides the most value, rather than throughout the entire workflow. Deterministic sections are maintained for tasks where predictability is sufficient, reducing overall computational overhead and execution time while still achieving flexibility where needed.
Solution Approach 2:
The patent performs preliminary actions by pre-training ML models offline and caching predictions where applicable. The system prepares probability models in advance and uses efficient inference techniques during workflow execution, reducing the computational burden and execution time during actual workflow runs.
Data Source
AI summary
Automatic activation and configuration of robotic process automation (RPA) workflows using machine learning (ML) is disclosed. One or more parts of an RPA workflow may be turned on or off based on one or more probabilistic ML models. RPA robots may be configured to modify parameters, determine how much of a certain resource to provide, determine more optimal thresholds, etc. Such RPA workflows implementing ML may thus be hybrids of both deterministic and probabilistic logic, and may learn and improve over time by retraining the ML models, adjusting the confidence thresholds, using local/global confidence thresholds, providing or adjusting modifiers for the local confidence thresholds, implement a supervisor system that monitors ML model performance, etc.