Dimensionality Reduction for Neural Network Descriptor Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
When training neural network models using a full set of descriptors, the training time becomes excessively long and inefficient, especially when the number of descriptors is large, and many descriptors contribute little to the training process.
Innovation Solution
Performing multiple dimensionality reduction iterations on the descriptor set until a valid quantity value N is reached, where the accuracy of the iterations remains stable and the number of descriptors remains approximately unchanged, and then determining core descriptors based on occurrence frequencies to output as attribute data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a full set of descriptors is used for training the neural network model, then the training completeness is improved, but the training time becomes excessively long and efficiency decreases
Solution Approach 1:
The patent extracts and removes redundant descriptors from the full descriptor set through dimensionality reduction iterations. By identifying and eliminating descriptors that contribute minimally to training effectiveness, the system retains only the essential descriptors needed for model training, thereby reducing training time while maintaining training completeness.
Solution Approach 2:
The patent applies partial action by using a subset of descriptors rather than the full set. Through iterative dimensionality reduction, the system determines the minimum necessary number of descriptors (N) required to achieve acceptable training accuracy, avoiding the excessive use of all available descriptors and thus improving training efficiency.
2Productivity
If the number of descriptors is reduced to improve training efficiency, then the training time is shortened, but the training effect may be impacted
Solution Approach 1:
The patent implements feedback through iterative dimensionality reduction processes. In each iteration, the system evaluates the training accuracy achieved with the current descriptor set and uses this feedback to determine whether further reduction is appropriate. The iteration continues until the improvement in training efficiency outweighs the potential loss in training effect, achieving an optimal balance.
Solution Approach 2:
The patent applies dynamics by making the descriptor set adaptive and adjustable. Rather than using a fixed full set or a fixed reduced set, the system dynamically adjusts the number and composition of descriptors through iterative reduction, allowing the descriptor set to evolve based on performance feedback and computational constraints.
Data Source
AI summary
A method and a system for determining attribute data that effectively describes a target object are provided. The method includes: obtaining a descriptor set that describes attributes of a target object, where the descriptor set includes K descriptors; performing a plurality of dimensionality reduction iterations on the descriptor set to reduce a quantity of descriptors in the descriptor set, until the plurality of dimensionality reduction iterations meet preset stop information; determining, from descriptors obtained by the plurality of dimensionality reduction iterations, descriptors whose occurrence frequencies meet a preset condition as core descriptors; and outputting the core descriptors and N as attribute data to improve validity and reliability of the attribute data used to describe the target object.