Multiclass SVM Reduced-Set Vector Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Supervised learning with multiclass support vector machines (SVMs) faces challenges in classification performance due to increased processing time with larger training datasets, which complicates real-time implementations.
Innovation Solution
Converting multiclass SVMs into binary problems, applying reduced-set methods like Burges and Gaussian reduced-set vector methods, and combining results into joint lists for re-training with reduction factors to optimize kernel functions and vector quantities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If larger training datasets are used to improve classification performance, then classification accuracy is improved, but processing time increases
Solution Approach 1:
The patent extracts and removes redundant support vectors from the training dataset through reduced-set methods. By identifying and eliminating vectors that do not contribute significantly to the decision boundary, the system maintains classification accuracy while reducing the dataset size, thereby decreasing processing time for real-time applications.
Solution Approach 2:
The patent applies parameter changes by modifying the reduction factor (RF) parameter to control the degree of dataset reduction. By adjusting this parameter, the system can optimize the balance between classification accuracy and processing speed, allowing flexible adaptation to different real-time application requirements.
2Productivity
If multiclass SVM is converted to binary problems and reduced-set methods are applied, then processing time is reduced, but classification performance may deteriorate
Solution Approach 1:
The patent applies preliminary action by pre-processing the multiclass SVM problem into binary problems and performing reduced-set methods before final classification. This preliminary reduction creates a streamlined dataset that maintains essential classification information while removing redundancy, enabling faster real-time processing without significant accuracy loss.
Solution Approach 2:
The system uses feedback mechanisms to evaluate classification performance after reduction and iteratively adjusts the reduction factor parameter. This feedback loop ensures that the reduced dataset maintains adequate classification accuracy while achieving the desired processing speed for real-time applications.
Data Source
AI summary
Techniques are disclosed for improving classification performance in supervised learning. In accordance with some embodiments, a multiclass support vector machine (SVM) having three or more classes may be converted to a plurality of binary problems that then may be reduced via one or more reduced-set methods. The resultant reduced-set (RS) vectors may be combined together in one or more joint lists, along with the original support vectors (SVs) of the different binary classes. Each binary problem may be re-trained using the joint list(s) by applying a reduction factor (RF) parameter to reduce the total quantity of RS vectors. In re-training, different kernel methods can be combined, in accordance with some embodiments. Reduction may be performed until desired classification performance is achieved. The disclosed techniques can be used, for example, to improve classification speed, accuracy, class prioritization, or a combination thereof, in the SVM training phase, in accordance with some embodiments.