State Determination Apparatus Using Semi-Supervised Learning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing techniques for determining whether a target is in a specific state, such as agitation, lack a method for accurately setting a threshold for the agitation score, leading to inaccuracies in state determination.

Innovation Solution

A state determination apparatus and method that utilize a calculation model generated by semi-supervised learning to calculate a score indicative of the target's state and a prediction model generated by supervised learning to decide a threshold, allowing for accurate comparison to determine the target's state.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a threshold is set for the agitation score to determine whether a target is in a specific state, then the determination can be made, but the accuracy of state determination deteriorates due to lack of guidance on threshold setting

Engineering Contradiction:

Improvestate determination accuracyVSAvoidthreshold setting guidance

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

The patent introduces an intermediary component (threshold determination unit) that uses prediction models to objectively determine the threshold value. This intermediary translates the raw agitation scores into meaningful decision boundaries by comparing predicted values against actual labeled data, thereby resolving the information loss about how to set appropriate thresholds.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements feedback mechanisms where prediction models are trained using labeled data to learn the relationship between agitation scores and actual states. The models continuously refine threshold determination by feedback from labeled examples, enabling accurate threshold setting without requiring manual specification of threshold values.

Inventive Principle:

Principle #23Feedback

2Measurement precision

If supervised learning is used to generate prediction models for threshold determination, then determination accuracy improves, but the requirement for labeled training data increases

Engineering Contradiction:

Improvethreshold determination accuracyVSAvoidlabeled training data

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent applies preliminary action by pre-training prediction models on available labeled data before actual state determination. The models are prepared in advance to learn the mapping between agitation scores and states, so that when new data arrives, the threshold determination can be performed accurately without requiring additional labeled data for each new determination task.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20240054362A1State determination apparatus using machine learning model, determination method, and storage medium

Publication Date: 2024.02.15 NEC CORP

AI summary

In order to achieve an object to determine, with high accuracy, that a target is in a specific state, a state determination apparatus includes: a calculation section that calculates, on the basis of data obtained from the target, a score indicative of a degree to which the target is in the specific state; a decision section that decides a threshold on the basis of the data; and a determination section that determines, by comparing the score and the threshold, whether the target is in the specific state.