Uncertainty-Based Data Collection Area Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data collection methods struggle to achieve consistent performance across various data areas due to the difficulty in setting effective data collection areas, especially when dealing with uncertainty in data accuracy and availability.
Innovation Solution
A method using uncertainty information to predict the areas of environmental information needed to be collected, involving the use of neural network models to output episodes, measure uncertainty, and calculate the average uncertainty to determine the required collection area.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data collection area is set intuitively or based on conventional accuracy, then data collection can be performed, but consistent performance cannot be achieved across various data areas
Solution Approach 1:
The patent changes the parameter used for setting data collection area from intuitive/conventional methods to uncertainty-based parameters. By calculating uncertainty values for different data areas and using these as the basis for setting collection areas, the system achieves consistent performance across various data areas while providing an objective, quantifiable method for area determination.
2Quantity of substance
If data collection area is set by counting the number of data, then data quantity can be determined, but the difficulty of learning varies by data area making consistent area setting difficult
Solution Approach 1:
The patent transitions from using data quantity (number of data points) as the sole criterion to using uncertainty values as the primary parameter for setting data collection areas. This allows the system to account for varying learning difficulties across different data areas by objectively measuring uncertainty, thereby achieving more consistent and reliable area determination regardless of data quantity.
3Measurement precision
If conventional accuracy-based method is used to set data collection area, then data collection can proceed, but data outside collection range cannot be known as ground truth
Solution Approach 1:
The patent applies preliminary action by calculating uncertainty values for all potential data areas before setting the final data collection area. This preliminary uncertainty assessment allows the system to identify which areas lack ground truth and prioritize them for collection, ensuring that ground truth availability is maintained while using accuracy-based methods.
4Productivity
If uncertainty-based prediction is used to determine data collection area, then targeted and efficient data collection is achieved, but additional computational processing is required
Solution Approach 1:
The patent applies preliminary action by performing uncertainty calculations in advance before actual data collection. This preliminary processing identifies the most critical data areas that need collection, allowing the system to achieve targeted and efficient data collection while minimizing redundant computational processing during the actual collection phase.
Data Source
AI summary
Disclosed is a method for predicting areas of environmental information needed to be collected, which is performed by one or more processors of a computing device. The method may include: outputting one or more episodes based on environmental information; measuring uncertainty for each of the one or more episodes; and predicting an area of the environmental information needed to be collected based on the measured uncertainty.