Problem
Research questions and friction points this paper is trying to address.
license plate recognition
data augmentation
instance-dependent label noise
instance selection
GERIS: A Game-Theoretic Framework for Filtering Instance-Dependent Label Noise in License Plate Data Augmentation
🤖 AI Summary
This study addresses the issues of noise-induced label dependency and model performance degradation in license plate data augmentation by proposing GERIS, a game-theoretic framework. This method integrates non-cooperative game theory, similarity metrics, and model reliability assessment to select high-quality synthetic samples while filtering out low-quality instances through a competitive mechanism. By doing so, it overcomes the lack of transparency inherent in conventional black-box approaches, achieving theoretically grounded and interpretable data filtering. Experimental results demonstrate that GERIS significantly outperforms existing instance selection methods in both classification accuracy and robustness, effectively enhancing the overall performance of recognition systems.
📝 Abstract
In this paper, we propose GERIS, a game-theoretic framework for instance selection in the data augmentation phase of license plate recognition systems. During augmentation, synthetic license plate images are generated and transformed using stochastic noise to simulate real-world conditions. However, certain noise configurations lead to highly distorted, unreadable images that degrade model performance by introducing instance-dependent label noise. GERIS formulates a non-cooperative game in which each noise vector competes for inclusion in the training set based on its similarity to labeled data and its contribution to model reliability. By identifying and pruning low-quality instances, GERIS improves the overall quality of the augmented dataset. Unlike traditional black-box learning methods, GERIS offers a transparent, theoretically grounded mechanism for data filtering. Experimental results demonstrate that GERIS outperforms existing instance selection methods in terms of classification accuracy and robustness.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Game-Theoretic Framework
Instance Selection
Label Noise Filtering
Data Augmentation
License Plate Recognition
🔎 Similar Papers
No similar papers found.
