Factor Inference Device Using Non-Supervised Deep Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies face challenges in estimating factors contributing to outcomes when the causal relationship is unknown, particularly in complex scenarios involving humans or machines, as they require teaching data that cannot be created for machine learning.
Innovation Solution
A factor estimation system and method utilizing deep learning with non-supervised learning to classify data and determine causal relationships, acquiring the ability to estimate factors by analyzing large volumes of data from various sources, including IoT sensors and wearable devices, to identify patterns and transitions between states and conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If deep learning with non-supervised learning is used to classify data and determine causal relationships, then the ability to estimate factors without pre-defined teaching data is improved, but the complexity of the learning system and data processing increases
Solution Approach 1:
The learning system is divided into multiple specialized components: a state classification learning device that classifies object states, a condition classification learning device that classifies conditions, and an outlier learning device that identifies exceptional patterns. This segmentation allows each component to focus on specific aspects of factor estimation, improving overall adaptability while managing complexity through modular design.
Solution Approach 2:
The system introduces intermediate processing layers including a state-information extraction device, condition-information extraction device, and learning data generating device. These intermediaries transform raw data into structured formats suitable for non-supervised learning, enabling factor estimation without pre-defined teaching data while organizing the complex processing pipeline into manageable stages.
2Measurement precision
If large volumes of data from various sources are analyzed to identify patterns and transitions, then the accuracy of factor estimation is improved, but the time and computational resources required increase
Solution Approach 1:
The system performs preliminary classification of states and conditions before conducting full factor estimation. By pre-organizing data into classified categories using the state classification learning device and condition classification learning device, the system reduces the computational burden during actual estimation, improving accuracy while reducing analysis time.
Solution Approach 2:
The outlier learning device focuses computational resources on identifying exceptional patterns and anomalies rather than processing all data uniformly. This partial action approach concentrates computational power on critical deviations that most impact factor estimation accuracy, reducing overall processing time while maintaining high precision.
3Extent of automation
If the system acquires the ability to classify data and determine causal relationships through non-supervised learning, then the reduction of human intervention is improved, but the difficulty of implementing and maintaining the learning models increases
Solution Approach 1:
The state classification learning device, condition classification learning device, and outlier learning device operate autonomously through non-supervised learning, continuously improving their classification abilities without human intervention. The system self-adjusts to new patterns and causal relationships, reducing the need for manual model updates while maintaining implementation feasibility through standardized learning algorithms.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A factor estimation device according to a mode of the present invention is configured to receive information pertaining to objects, to extract state information from the information received, to identify a predetermined state pertaining to a first object from among the objects, to receive state information extracted that corresponds to the predetermined state and classify the aforementioned predetermined state, to extract condition information from the information received, to identify the condition up until the predetermined state, and to receive condition information that is output by the condition-information extraction unit and corresponds to the condition identified and classify the aforementioned condition identified. Subsequently, the factor estimation device is configured to estimate the condition that may result in the predetermined state on the basis of the result of classifying the predetermined state and the result of classifying the identified condition.