State Estimation Model Training Using Dynamic Reference Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating a user's psychological state from behavior data fail to accurately account for time-series fluctuations and individual differences in self-evaluation, leading to inconsistent results and reliance on arbitrary reference points.
Innovation Solution
A training apparatus that extracts feature vector data from behavior data, calculates difference vectors based on past data points, and trains a state estimation model using these vectors to provide accurate psychological state estimation, incorporating a self-attention mechanism to assess importance at each time point.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data is separated and processed on a daily basis using related-art methods, then processing simplicity is maintained, but accuracy in estimating user's psychological state deteriorates because time-series fluctuations and individual differences in self-evaluation are not considered
Solution Approach 1:
The patent applies dynamics by making the reference point for psychological state estimation dynamic rather than fixed. The system automatically determines appropriate reference points based on time-series data patterns, allowing the estimation to adapt to individual user behaviors and temporal variations. This resolves the contradiction by improving accuracy through dynamic adaptation while maintaining processing feasibility through automated reference point selection.
Solution Approach 2:
The patent introduces a temporal dimension to the estimation process by incorporating time-series data and comparing current states with historical reference points. This dimensional expansion allows the system to capture time-dependent patterns and individual differences, improving estimation accuracy without requiring overly complex manual processing by adding structured temporal analysis.
2Reliability
If arbitrary reference points are used for self-evaluation comparison, then processing complexity is reduced, but reliability of estimation results deteriorates due to inconsistency in reference point selection
Solution Approach 1:
The system applies self-service by automatically selecting and determining reference points based on the user's own historical data patterns. The algorithm identifies meaningful reference points from the time-series data without requiring external intervention or arbitrary manual selection. This ensures consistent and reliable reference point selection while maintaining processing efficiency through automation.
Solution Approach 2:
The patent implements feedback by using the determined reference points to continuously improve future estimations. The system learns from past estimation accuracy and adjusts reference point selection accordingly, creating a feedback loop that enhances reliability over time. This automated feedback mechanism ensures consistent reference point selection without increasing manual complexity.
3Loss of information
If traditional daily-based data processing is used, then processing speed is maintained, but information completeness deteriorates because individual differences and temporal patterns are not captured
Solution Approach 1:
The patent applies preliminary action by pre-processing the time-series data to identify and mark potential reference points before the actual estimation process. This preliminary analysis organizes the data in a way that preserves individual differences and temporal patterns while making the subsequent estimation process efficient. The reference points are determined in advance based on data patterns, reducing information loss without significantly impacting processing speed.
Solution Approach 2:
The patent segments the continuous time-series data into meaningful units by identifying discrete reference points that capture essential patterns. This segmentation preserves critical information about individual differences and temporal variations while making the data more manageable for processing. By dividing the data into relevant segments rather than processing it as a continuous stream, the system maintains both information completeness and processing efficiency.
Data Source
AI summary
A training apparatus includes a feature extraction unit that extracts feature vector data from behavior data of each date and time, a reference point extraction unit that performs, for each date and time, processing of calculating a difference between feature vector data of certain date and time and each of one or more pieces of feature vector data in past within a predetermined period from the date and time, and extracting one or more pieces of difference vector data corresponding to the feature vector data of the date and time, and a state estimation model training unit that trains a state estimation model using feature vector data of each date and time, difference vector data, and state information.