Combination of linear classifiers using score function -- analysis of possible combination strategies
This work addresses an incremental improvement in classifier combination methods for machine learning practitioners, with no broad impact indicated.
The paper tackled the problem of combining linear classifiers using score functions based on distance from decision boundaries, testing two score functions and four combination strategies, and found that simple average and trimmed average strategies performed best in experiments compared to majority voting and model averaging across seven quality criteria.
In this work, we addressed the issue of combining linear classifiers using their score functions. The value of the scoring function depends on the distance from the decision boundary. Two score functions have been tested and four different combination strategies were investigated. During the experimental study, the proposed approach was applied to the heterogeneous ensemble and it was compared to two reference methods -- majority voting and model averaging respectively. The comparison was made in terms of seven different quality criteria. The result shows that combination strategies based on simple average, and trimmed average are the best combination strategies of the geometrical combination.