ML Resource Scheduling Engine for Predictive Utilization Control

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing computer systems face challenges in optimizing resource utilization due to finite resources and varying availability over time, necessitating improved scheduling methods to enhance efficiency.

Innovation Solution

A machine-learning (ML)-based engine that compiles historical resource utilization data to determine relevant parameters and weighting, calculates distance values for potential runtimes, and schedules computing jobs to optimize resource utilization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional scheduling methods are used, then system simplicity is maintained, but resource utilization efficiency deteriorates

Engineering Contradiction:

Improveresource utilization efficiencyVSAvoidscheduling system complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent introduces an ML-based intermediary engine that acts as a mediator between job scheduling requests and resource allocation. This engine compiles historical resource utilization data and uses ML models to predict future resource needs, thereby improving scheduling efficiency without requiring complex manual intervention or system restructuring

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system performs preliminary actions by compiling and analyzing historical resource utilization data before actual scheduling decisions are made. The ML model predicts future resource requirements in advance, allowing the system to proactively optimize scheduling decisions rather than reacting to resource constraints in real-time

Inventive Principle:

Principle #10Preliminary action

2Productivity

If resources are allocated dynamically, then resource utilization is improved, but prediction accuracy requirements increase

Engineering Contradiction:

Improveresource utilizationVSAvoidresource prediction accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements a feedback mechanism where the ML model continuously learns from actual resource utilization outcomes. The system compiles historical data including predicted versus actual resource usage, uses this feedback to refine prediction accuracy, and dynamically adjusts scheduling decisions based on improved predictions, thereby resolving the tension between dynamic allocation and prediction accuracy requirements

Inventive Principle:

Principle #23Feedback

3Measurement precision

If historical data is extensively analyzed, then prediction accuracy improves, but processing time increases

Engineering Contradiction:

Improveprediction accuracyVSAvoiddata processing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary action by pre-compiling and organizing historical resource utilization data into structured formats before prediction is needed. The ML model is trained in advance on this compiled data, so that during actual scheduling operations, predictions can be generated quickly without re-processing the entire historical dataset, thus resolving the conflict between comprehensive data analysis and processing time

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12585501B2Machine-learning (ML)-based resource utilization prediction and management engine

Publication Date: 2026.03.24 BANK OF AMERICA CORP

AI summary

Systems and methods for optimizing resource utilization in a computing system are provided. Methods include compiling training data, running a machine-learning (ML) model using the training data to determine a set of resource parameters, receiving a request to schedule the computing job on the computing system, computing, for each of a plurality of potential runtimes, an amalgam distance value, selecting the potential runtime with the optimal distance value for scheduling the computing job, and generating a job execution schedule based on the selected potential runtime.