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.

翻译：本文重新审视了深度互学习这一简单而有效的计算范式。我们提出使用Rényi散度替代KL散度，因其更具灵活性和可调性，以改进原始DML方法。该改进方案能在有限增加复杂度的前提下，持续提升原始DML的性能。我们通过理论分析证明了所提范式的收敛特性：在非凸优化任务的最坏情况下，恒定学习率的随机梯度下降算法能以$\mathcal{O}(1)$偏差收敛。这意味着学习过程将抵达邻近的局部最优解，并在有界范围内持续搜索，这可能有助于缓解过拟合问题。最终，大量实证结果表明，结合DML与Rényi散度的策略能有效提升模型泛化能力。