Maximum Entropy Regularization and Chinese Text Recognition

TL;DR

This paper introduces a maximum entropy regularization method for Chinese text recognition, improving model generalization and robustness by adding a negative entropy term to the loss function without extra parameters.

Contribution

The paper proposes a simple yet effective maximum entropy regularization technique for Chinese text recognition, with theoretical analysis and empirical validation demonstrating its benefits.

Findings

Improved recognition accuracy across Chinese character and text line recognition tasks.

Enhanced model robustness and generalization in fine-grained image classification.

Theoretical insights into the convergence and influence of the regularization.

Abstract

Chinese text recognition is more challenging than Latin text due to the large amount of fine-grained Chinese characters and the great imbalance over classes, which causes a serious overfitting problem. We propose to apply Maximum Entropy Regularization to regularize the training process, which is to simply add a negative entropy term to the canonical cross-entropy loss without any additional parameters and modification of a model. We theoretically give the convergence probability distribution and analyze how the regularization influence the learning process. Experiments on Chinese character recognition, Chinese text line recognition and fine-grained image classification achieve consistent improvement, proving that the regularization is beneficial to generalization and robustness of a recognition model.