Learning Tree Output Node Selection Using Reliability Index
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In learning methods using a tree structure, lower layers have smaller state spaces with less learned data, making them more susceptible to noise, leading to unreliable prediction outputs, especially when learning has not proceeded sufficiently.
Innovation Solution
An information processing device that generates prediction outputs by specifying input nodes based on input data and using reliability indices to select a higher reliable node for prediction, with error calculations and forgetting coefficients to adjust the influence of new and existing data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If learning is performed using a tree structure with hierarchically divided state spaces, then computation speed is improved and memory usage is reduced, but prediction reliability deteriorates when learning has not proceeded sufficiently due to noise susceptibility in lower layers
Solution Approach 1:
The patent dynamically selects the output node based on the reliability index rather than always using the end node. The system adjusts which node serves as the output node depending on the learned model's reliability, allowing it to adapt between using deeper nodes (for better accuracy when reliable) and higher nodes (for noise resistance when unreliable), thus resolving the contradiction between computation speed and prediction reliability
Solution Approach 2:
The patent changes the parameter of output node selection from a fixed end-node approach to a dynamic selection based on the reliability index. By introducing the reliability index as a selection criterion and allowing the output node position to vary based on this parameter, the system can optimize between speed and reliability depending on the learning state
2Measurement precision
If lower layer nodes are used for prediction, then prediction accuracy is improved through more specific state spaces, but noise susceptibility increases leading to reduced reliability when data is insufficient
Solution Approach 1:
The patent uses the reliability index as feedback to determine output node selection. The reliability index, calculated from the learned model's performance, provides feedback that guides whether to use lower layer nodes (for accuracy) or higher layer nodes (for noise resistance), thus resolving the contradiction between prediction accuracy and noise resistance
Solution Approach 2:
The system dynamically adjusts the output node selection based on the reliability index, allowing it to move between higher and lower layer nodes depending on the learning state and noise conditions, thereby balancing prediction accuracy and noise resistance
Data Source
AI summary
An information processing device generates a prediction output corresponding to input data. The information processing device includes input-node specification processor circuitry, based on the input data, configured to specify input nodes corresponding to the input data and each located on a corresponding one of layers from beginning to end of the learning tree structured, reliability-index acquisition processor circuitry configured to acquire a reliability index obtained through the predetermined learning processing and indicating prediction accuracy, output-node specification processor circuitry, based on the reliability index acquired by the reliability-index acquisition processor circuitry configured to specify, from the input nodes corresponding to the input data, an output node that is the basis of the generation of a prediction output, and prediction-output generation processor circuitry configured to generate a prediction output.