Predictive Workflow Models With Reward-Guided Hyperparameter Search

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Software applications struggle to present workflow sequences efficiently and effectively to users due to varying user responses to different sequences, lacking sufficient historical data for customization, and relying on descriptive models that react to past actions rather than proactively optimizing user experience.

Innovation Solution

Training predictive models using hyperparameter searching to identify optimal workflow sequences that maximize a reward metric by generating training data from random user interactions, evaluating multiple models with different hyperparameter sets, and deploying the most optimized model to dynamically generate user interfaces based on user-specific prioritization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If descriptive models are used to react to past user actions, then the system can process historical data, but it cannot proactively optimize user experience or predict optimal workflow sequences

Engineering Contradiction:

Improveuser experience optimizationVSAvoidreactive vs proactive behavior

Core Design Contradiction:

Adaptability or versatilityVSExtent of automation

Solution Approach 1:

The system performs preliminary actions by predicting optimal workflow sequences before users actually execute them. The predictive model analyzes historical data and proactively determines the best sequence of workflow steps to present to users, rather than merely reacting to past actions. This allows the system to prepare and present optimized workflows in advance, improving user experience by anticipating user needs.

Inventive Principle:

Principle #10Preliminary action

2Productivity

If multiple workflow sequences are presented to users, then user efficiency can be improved through customization, but the number of possible sequences increases factorially with workflow complexity

Engineering Contradiction:

Improveuser efficiencyVSAvoidnumber of workflow sequences

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The system changes parameters by using predictive models to transform the exponential space of possible workflow sequences into a manageable set of optimized sequences. Instead of presenting all factorial possibilities, the model predicts and selects only the most relevant sequences based on user characteristics and historical data, effectively reducing the complexity while maintaining high user efficiency.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system extracts only the essential information needed for optimization from the vast space of possible workflow sequences. By analyzing historical data and user patterns, the system extracts key insights to generate a limited set of high-quality sequences, discarding unnecessary complexity while preserving the most valuable workflow variations for user presentation.

Inventive Principle:

Principle #2Taking out (Extraction)

3Adaptability or versatility

If historical data is collected for customization, then user-specific workflow optimization is possible, but insufficient historical data limits the ability to provide accurate customization

Engineering Contradiction:

Improveuser-specific customizationVSAvoidamount of historical data

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The system achieves universality by building a predictive model that can generalize from limited historical data to provide accurate customizations. The model is designed to learn patterns from available data and apply them across different users and workflows, making the system effective even with insufficient historical data for any single user by leveraging universal patterns from the broader population.

Inventive Principle:

Principle #6Universality (Multi-functionality)

4Productivity

If workflow sequences are customized for each user, then user efficiency improves, but computing resources are consumed presenting and processing multiple sequences

Engineering Contradiction:

Improveuser efficiencyVSAvoidcomputing resource consumption

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The system performs preliminary computation by pre-calculating and storing optimized workflow sequences based on predictive models before actual user interaction. This allows the system to present pre-processed, optimized sequences rather than computing them in real-time during user interaction, significantly reducing computing resource consumption while maintaining high user efficiency.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20250371437A1Training predictive models based on reward signals and hyperparameter searching

Publication Date: 2025.12.04 INTUIT INC

AI summary

Aspects of the present disclosure provide techniques for training and using machine learning models to predict and present an optimal workflow to a user of a software application. An example method generally includes generating a training data set including a plurality of exemplars including features associated a user of a software application, a sequence of workflow steps presented to the user of the software application, and a reward metric. A plurality of hyperparameter sets for training a plurality of predictive models is generated. The plurality of predictive models are trained based on the plurality of hyperparameter sets. A hyperparameter set from the plurality of hyperparameter sets is selected based on performance metrics for each of the plurality of predictive models. A machine learning model is trained based on the selected hyperparameter set and the training data set, and the trained machine learning model is deployed.