Neural Network Data Conversion Rules for Consistent Variable Ordering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In neural network-based learning methods, the accuracy of classifying relationships between people and things is compromised when different conversion rules are applied to similar relationship data, leading to inconsistent ordering of common variables, which affects the association between data and deteriorates learning and classification accuracy.
Innovation Solution
A method where conversion rules for relationship data are determined such that the correspondence relationships of common variables are made consistent across different data sets, ensuring that the ordering of input data to the neural network is uniform, thereby maintaining the association between relationship data and improving learning accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If different conversion rules are applied to relationship data, then flexibility in data processing is improved, but consistency of correspondence relationships deteriorates
Solution Approach 1:
The patent applies homogeneity by making the conversion rules for common variables consistent across different relationship data sets. Specifically, when multiple relationship data share a common variable (e.g., the same person or object), the patent ensures they are converted to the same ordinal position according to a unified collation pattern, rather than allowing each data set to have its own independent conversion rules.
2Stability of the object's composition
If conversion rules are standardized across relationship data, then consistency of correspondence relationships is improved, but adaptability to different data structures deteriorates
Solution Approach 1:
The patent segments the conversion process into two distinct parts: (1) determination of collation patterns for common variables that ensures consistency, and (2) independent conversion of unique variables that maintains adaptability. This segmentation allows the system to apply uniform rules where needed while preserving flexibility for data-specific characteristics.
3Measurement precision
If collation patterns are updated using error back propagation, then classification accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-determining collation patterns based on the structural characteristics of relationship data before the main classification process. By establishing the ordering of common variables in advance according to their semantic relationships, the system reduces the computational burden during error back propagation, as the neural network only needs to optimize the conversion of unique variables rather than determining the entire data structure from scratch.
Data Source
AI summary
A learning method includes: acquiring input data and correct answer information, the input data including a set of multiple pieces of relationship data in which relationships between variables are recorded respectively; determining conversion rule corresponding to each of the multiple pieces of relationship data such that relationships before and after a conversion of a common variable commonly in the multiple pieces of relationship data are the same, when converting a variable value in each of the multiple pieces of relationship data into converted data rearranging the variable values in an order of input; converting each of the multiple pieces of relationship data into a multiple pieces of the converted data according to each corresponding conversion rule; and inputting a set of the multiple pieces of converted data to the neural network and causing the neural network to learn a learning model based on the correct answer information.