Rényi Divergence Deep Mutual Learning
Preprint
Abstract
This paper revisits Deep Mutual Learning (DML), a simple yet effective computing paradigm. We propose using R\'{e}nyi divergence instead of the KL divergence, which is more flexible and tunable, to improve vanilla DML. This modification is able to consistently improve performance over vanilla DML with limited additional complexity. The convergence properties of the proposed paradigm are analyzed theoretically, and Stochastic Gradient Descent with a constant learning rate is shown to converge with $\mathcal{O}(1)$-bias in the worst case scenario for nonconvex optimization tasks. That is, learning will reach nearby local optima but continue searching within a bounded scope, which may help mitigate overfitting. Finally, our extensive empirical results demonstrate the advantage of combining DML and R\'{e}nyi divergence, leading to further improvement in model generalization.Keywords
All Related Versions
This publication has 15 references indexed in Scilit:
- DCML: Deep contrastive mutual learning for COVID-19 recognitionBiomedical Signal Processing and Control, 2022
- Deep mutual learning for visual object trackingPattern Recognition, 2021
- MutualNet: Adaptive ConvNet via Mutual Learning from Network Width and Resolution2020
- Analysis of biased stochastic gradient descent using sequential semidefinite programsMathematical Programming, 2020
- Diversified Mutual Learning for Deep Metric Learning2020
- Deep Mutual Learning2018
- Improved Security Proofs in Lattice-Based Cryptography: Using the Rényi Divergence Rather than the Statistical DistanceJournal of Cryptology, 2017
- Deep Residual Learning for Image Recognition2016
- Going deeper with convolutions2015
- Describing Textures in the Wild2014