Mixup is a widely adopted strategy for training deep networks, where additional samples are augmented by interpolating inputs and labels of training pairs. Mixup has shown to improve classification performance, network calibration, and out-of-distribution generalisation. While effective, a cornerstone of Mixup, namely that networks learn linear behaviour patterns between classes, is only indirectly enforced since the output interpolation is performed at the probability level. This paper seeks to address this limitation by mixing the classifiers directly instead of mixing the labels for each mixed pair. We propose to define the target of each augmented sample as a uniquely new classifier, whose parameters are a linear interpolation of the classifier vectors of the input pair. The space of all possible classifiers is continuous and spans all interpolations between classifier pairs. To make optimisation tractable, we propose a dual-contrastive Infinite Class Mixup loss, where we contrast the classifier of a mixed pair to both the classifiers and the predicted outputs of other mixed pairs in a batch. Infinite Class Mixup is generic in nature and applies to many variants of Mixup. Empirically, we show that it outperforms standard Mixup and variants such as RegMixup and Remix on balanced, long-tailed, and data-constrained benchmarks, highlighting its broad applicability.

翻译：混合是一种广泛采用的深度网络训练策略，通过插值训练对的输入和标签来增强样本。混合已被证明能提升分类性能、网络校准和分布外泛化能力。尽管有效，混合的一个核心基石——即网络学习类别间的线性行为模式——仅通过间接方式实现，因为输出插值是在概率层面进行的。本文旨在通过直接混合分类器而非混合每个混合对的标签来解决这一局限。我们提出将每个增强样本的目标定义为一个全新的独特分类器，其参数是输入对分类器向量的线性插值。所有可能分类器的空间是连续的，并覆盖分类器对之间的所有插值。为使优化可行，我们提出一个双重对比的无限类别混合损失，其中我们将混合对的分类器与批次中其他混合对的分类器及预测输出进行对比。无限类别混合本质上是通用的，并适用于混合的多种变体。实验表明，它在平衡、长尾和数据受限基准上优于标准混合及其变体如RegMixup和Remix，凸显了其广泛的适用性。