Be Your Own Best Competitor! Multi-Branched Adversarial Knowledge Transfer

TL;DR

This paper introduces a multi-branched adversarial knowledge transfer method that enhances lightweight neural networks' performance through self-distillation and adversarial training, improving accuracy without extra inference costs.

Contribution

It proposes a novel ensemble self-distillation approach with adversarial learning for compact models, outperforming existing methods in accuracy and efficiency.

Findings

Outperforms primary models in accuracy at same parameter count

Effective in both image classification and encoder-decoder architectures

Achieves significant improvements over previous self-distillation techniques

Abstract

Deep neural network architectures have attained remarkable improvements in scene understanding tasks. Utilizing an efficient model is one of the most important constraints for limited-resource devices. Recently, several compression methods have been proposed to diminish the heavy computational burden and memory consumption. Among them, the pruning and quantizing methods exhibit a critical drop in performances by compressing the model parameters. While the knowledge distillation methods improve the performance of compact models by focusing on training lightweight networks with the supervision of cumbersome networks. In the proposed method, the knowledge distillation has been performed within the network by constructing multiple branches over the primary stream of the model, known as the self-distillation method. Therefore, the ensemble of sub-neural network models has been proposed to transfer the knowledge among themselves with the knowledge distillation policies as well as an adversarial learning strategy. Hence, The proposed ensemble of sub-models is trained against a discriminator model adversarially. Besides, their knowledge is transferred within the ensemble by four different loss functions. The proposed method has been devoted to both lightweight image classification and encoder-decoder architectures to boost the performance of small and compact models without incurring extra computational overhead at the inference process. Extensive experimental results on the main challenging datasets show that the proposed network outperforms the primary model in terms of accuracy at the same number of parameters and computational cost. The obtained results show that the proposed model has achieved significant improvement over earlier ideas of self-distillation methods. The effectiveness of the proposed models has also been illustrated in the encoder-decoder model.