Neural Data Classification With Similarity Checks for Unknown Classes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing classifiers using convolutional neural networks may inaccurately classify data of unknown classes into known classes, leading to decreased classification accuracy.
Innovation Solution
A data classification device and method that utilizes a neural network-based model to calculate similarity between input data and learned data, determining whether the data belongs to known classes or unknown classes based on similarity standards, and displaying similar data for user verification, with evaluation of the model's accuracy through determination evaluation values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a classifier using a neural network is used to classify data, then classification speed and automation are improved, but classification accuracy deteriorates when unknown class data is misclassified as known class
Solution Approach 1:
The patent introduces a similarity calculation unit as an intermediary between the classification unit and the determination unit. This intermediary calculates the similarity between input data and learned data, providing additional information that helps the determination unit accurately identify unknown class data without slowing down the overall classification process.
Solution Approach 2:
The patent segments the classification process into distinct functional units: a classification unit that performs initial classification, a similarity calculation unit that computes similarity metrics, and a determination unit that makes final decisions. This segmentation allows each unit to specialize in its function while working together to improve overall accuracy without sacrificing speed.
2Measurement precision
If similarity calculation is added to determine unknown classes, then classification accuracy is improved, but device complexity increases
Solution Approach 1:
The determination unit serves multiple functions: it receives classification results from the classification unit, evaluates similarity metrics from the similarity calculation unit, and makes final determination decisions. This multi-functionality reduces the need for separate dedicated components for each function, thereby managing complexity while improving accuracy.
3Reliability
If determination evaluation and model reconstruction alerts are implemented, then reliability is improved, but loss of time increases due to additional evaluation steps
Solution Approach 1:
The determination evaluation unit automatically monitors the performance of the learned model and generates alerts when reconstruction is needed, without requiring manual intervention. This self-service approach ensures reliable model monitoring while minimizing time loss by eliminating human review steps.
Data Source
AI summary
A data classification device includes: a data acquisition unit configured to acquire data to be classified; a classification unit configured to classify the data into one of a plurality of classes by using a learned model learned using a neural network; a similarity calculation unit that calculates the similarity between the data and learned data used to generate the learned model; and a determination unit configured to determine whether the data belongs to the classes on the basis of the similarity, and, when the data does not belong to any of the classes, determine that the data belongs to an unknown class.