Material Classification via Classifier Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional material classification methods rely heavily on training data and specific internal implementations of classifiers, making them unsophisticated and time-consuming, and often incompatible when different classifiers are trained on different data sets, leading to potential misclassifications and increased processing times.
Innovation Solution
Combining pre-existing classifiers based on different approaches without prior knowledge of their implementation or training data, using predesignated stored associations to classify materials into probability distributions, allowing for dynamic combination and modification of classifiers without relying on training data, and enabling accurate classification without a learning process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional classifiers are combined using fixed internal implementations, then classification accuracy can be improved, but the system becomes complex and difficult to maintain
Solution Approach 1:
The system segments the classification task by separating individual classifiers into independent modules, each with its own internal implementation. These modular classifiers can be combined through a simple association mechanism without exposing their internal complexities, thus maintaining accuracy while reducing system complexity.
Solution Approach 2:
An intermediary association structure is introduced between classifiers and material types. This association layer acts as a mediator that connects classifiers to their target material types without requiring knowledge of the classifiers' internal implementations, simplifying the overall system architecture.
2Measurement precision
If classifiers are trained on different data sets, then each classifier can be optimized for specific materials, but compatibility issues arise and processing time increases
Solution Approach 1:
The system changes the parameter of classifier association from data-dependent to predetermined. Instead of dynamically selecting classifiers based on training data compatibility, the system uses predetermined associations between classifiers and material types, eliminating compatibility checks and reducing processing time.
3Measurement precision
If training data is used to train classifiers, then classification performance can be optimized, but the system becomes dependent on potentially unreliable or insufficient training data
Solution Approach 1:
The system extracts the training data dependency from the classification process. By using predetermined associations between classifiers and material types rather than learning from training data, the system eliminates reliance on potentially unreliable or insufficient training data while maintaining classification performance.
4Adaptability or versatility
If a learning process is used to train classifiers, then the system can adapt to new data, but the process is time-consuming and cannot be dynamically adjusted
Solution Approach 1:
The system introduces dynamic adjustability through predetermined associations that can be modified without retraining. The association between classifiers and material types can be dynamically updated based on new information or requirements, providing adaptability without the time-consuming training process.
Data Source
AI summary
The present disclosure relates to classification of a material type of an unknown material. The material type is classified into a probability distribution of multiple predetermined material types by using a collection of plural predetermined material classifiers. Each material type of the multiple predetermined material types is associated with a corresponding best performing classifier from the collection of plural predetermined material classifiers by a predesignated stored association. The plural material classifiers are applied to the unknown material to obtain a list of candidate material types. A list of potential best performing classifiers is looked up using the list of candidate material types as a reference into the predesignated stored association. A respective probability that the unknown material belongs to a material type is assigned based on the list of potential best performing classifiers.