Machine Learning Dataset Consolidation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing database management systems face challenges in efficiently managing datasets due to duplicate data storage, which increases infrastructure costs and can lead to inaccurate data analysis.

Innovation Solution

A computing system that utilizes machine learning processes to detect data redundancies across multiple datasets, involving entity data processing, validation, and training of model architectures to identify data similarities and facilitate dataset consolidation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If duplicate datasets are stored to various databases and servers, then data availability for multiple teams is improved, but infrastructure costs increase

Engineering Contradiction:

Improvedata availabilityVSAvoidinfrastructure costs

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent consolidates multiple duplicate datasets into a single centralized dataset, merging redundant storage locations into one unified data repository that serves multiple teams, thereby reducing infrastructure costs while maintaining data availability

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The centralized dataset serves as a universal data source for multiple teams and projects, allowing one dataset to fulfill the data needs of various teams simultaneously, eliminating the need for separate duplicate datasets for each team

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Ease of operation

If duplicate datasets are stored to various databases and servers, then data access for different projects is improved, but data accuracy deteriorates

Engineering Contradiction:

Improvedata accessVSAvoiddata accuracy

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

By merging duplicate datasets into a single centralized dataset, the patent eliminates inconsistencies between duplicates, ensuring that all teams access the same accurate data while maintaining ease of access through the unified repository

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS20250181572A1Database and file management systems and methods for consolidating datasets

Publication Date: 2025.06.05 TRUIST BANK

AI summary

Systems and methods receive input(s) facilitating dataset management, the input(s) initiating a machine learning process configured to detect data redundancies of two or more datasets. Entity data stored to entity data storage location(s) are accessed and processed to conform with formatting requirements for the machine learning process. Validation is performed on the processed entity data, the validation ensuring the processed entity data satisfy the formatting requirements, where the validation produces training data that is inserted into an iterative training and testing loop. A model architecture is trained, based on weights and calculations, using the training data in the iterative training and testing loop to detect data redundancies, the training including predicting a target variable and iteratively adjusting the weights and calculations during each subsequent iteration to improve predictability of the target variable, where the model architecture is trained to identify data similarities among the two or more datasets.