Input Relevancy Condensation for Faster AI Event Prediction

Solution Overview

Problem

Existing AI systems require significant processing time and power due to large input data sets, which can reduce accuracy when data reduction is implemented, necessitating a system to maintain accuracy while reducing processing time and power consumption.

Innovation Solution

A system that reduces input data size by eliminating irrelevant data points through time windowing and association with interface channels or event characteristics, forming a condensed input data set for machine learning algorithms.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If additional input data is provided to an AI system, then the accuracy of the generated output increases, but the processing time and computing power consumption increase

Engineering Contradiction:

Improveaccuracy of outputVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent extracts and removes irrelevant or redundant data points from the input dataset before feeding it to the AI system. By identifying and eliminating data that does not contribute to prediction accuracy, the system reduces processing time while maintaining the accuracy achieved with full relevant data.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies different processing treatments to different portions of the input data based on their relevance and quality. High-value relevant data is retained and processed with full attention, while irrelevant data is removed, creating a non-uniform processing strategy that optimizes the balance between accuracy and efficiency.

Inventive Principle:

Principle #3Local quality

2Measurement precision

If additional input data is provided to an AI system, then the accuracy of the generated output increases, but the computing power consumption increases

Engineering Contradiction:

Improveaccuracy of outputVSAvoidcomputing power consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent extracts and removes irrelevant or redundant data points from the input dataset before feeding it to the AI system. By identifying and eliminating data that does not contribute to prediction accuracy, the system reduces computing power consumption while maintaining the accuracy achieved with full relevant data.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies different processing treatments to different portions of the input data based on their relevance and quality. High-value relevant data is retained and processed with full attention, while irrelevant data is removed, creating a non-uniform processing strategy that optimizes the balance between accuracy and energy efficiency.

Inventive Principle:

Principle #3Local quality

3Productivity

If the quantity of input data is reduced, then the processing time and computing power consumption decrease, but the accuracy of the determined characteristic reduces

Engineering Contradiction:

Improveprocessing efficiencyVSAvoidaccuracy of prediction

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent extracts and removes irrelevant or redundant data points from the input dataset before feeding it to the AI system. By identifying and eliminating data that does not contribute to prediction accuracy, the system reduces processing time while maintaining the accuracy achieved with full relevant data.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies different processing treatments to different portions of the input data based on their relevance and quality. High-value relevant data is retained and processed with full attention, while irrelevant data is removed, creating a non-uniform processing strategy that optimizes the balance between accuracy and efficiency.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12547891B2Real-time prediction of future events using integrated input relevancy

Publication Date: 2026.02.10 TRUIST BANK

AI summary

A system includes a computer to implement a front-end input condensation program and a back-end machine learning program. Steps of the front-end program include receive input data and time data indicative of previous events associated with users; determine interface channels associated with modes of interface with the users and/or previous event characteristics; associate previous event data with time windows; generate user window values for the combinations of users and time windows indicating the interface channels and previous event characteristics of data within the respective time windows; and form condensed input data without raw input data having a low association with respect to preceding the subsequent event. Steps of the back-end program include receive the condensed input data and use the condensed data to generate an inference related to the subsequent event such that a time required by the machine learning algorithm to generate the inference is reduced.