Confidence-Guided Neural Network Classification With Transfer Learning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional classification systems and methods are costly, cumbersome, and inefficient.

Innovation Solution

A method and system utilizing a pre-trained neural network to determine confidence levels associated with classes, adapt through transfer learning, and perform reclassification when confidence levels fall within or exceed evaluation bands, with options for machine or human intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional classification systems are used, then classification can be performed, but the system is costly and cumbersome

Engineering Contradiction:

Improveclassification accuracyVSAvoidsystem complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces conventional mechanical classification systems with a neural network-based system. The neural network automatically learns classification boundaries from training data, substituting manual feature engineering and threshold-based classification with adaptive, data-driven models that reduce system complexity while maintaining or improving accuracy

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the parameters of the classification system by using confidence levels as an additional dimension for decision-making. Instead of relying solely on class labels, the system incorporates confidence metrics to determine when reclassification is needed, allowing dynamic adjustment of classification behavior based on uncertainty levels

Inventive Principle:

Principle #35Parameter changes

2Productivity

If conventional classification systems are used, then classification can be performed, but the process is inefficient

Engineering Contradiction:

Improveclassification efficiencyVSAvoidprocessing time

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent implements preliminary action by pre-training neural networks on extensive datasets before deployment. The confidence level thresholds and evaluation bands are pre-established during the training phase, allowing the system to make rapid classification decisions during operation without requiring complex real-time computations

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent incorporates feedback mechanisms where classification results and confidence levels are continuously evaluated. When confidence levels fall within evaluation bands, the system triggers reclassification processes that feed back into the neural network, improving future classifications while maintaining efficient operation for high-confidence cases

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20250307647A1Method and system for classification

Publication Date: 2025.10.02 II VI DELAWARE INC

AI summary

The present disclosure may comprise a classification method and system, to data at a pre-trained neural network. The pre-trained neural network may determine one or more confidence levels based on the data. Each of the one or more confidence levels may be associated with a class and an evaluation band. If at least one of the determined one or more confidence levels falls within its associated evaluation band the determined one or more confidence levels and the data may be collected for reclassification and adapting the pre-trained neural network by transfer learning.