Distribution shifts between training and test data are all but inevitable over the lifecycle of a deployed model and lead to performance decay. Adapting the model can hopefully mitigate this drop in performance. Yet, adaptation is challenging since it must be unsupervised: we usually do not have access to any labeled data at test time. In this paper, we propose a probabilistic state-space model that can adapt a deployed model subjected to distribution drift. Our model learns the dynamics induced by distribution shifts on the last set of hidden features. Without requiring labels, we infer time-evolving class prototypes that serve as a dynamic classification head. Moreover, our approach is lightweight, modifying only the model's last linear layer. In experiments on real-world distribution shifts and synthetic corruptions, we demonstrate that our approach performs competitively with methods that require back-propagation and access to the model backbone. Our model especially excels in the case of small test batches - the most difficult setting.

翻译：在模型部署的生命周期中，训练数据与测试数据之间的分布偏移几乎不可避免，并会导致性能衰减。对模型进行适应有望缓解这种性能下降。然而，适应过程具有挑战性，因为它必须是无监督的：在测试时我们通常无法获得任何标注数据。本文提出一种概率状态空间模型，能够对经历分布漂移的已部署模型进行适应。我们的模型学习分布偏移在最后一组隐藏特征上引发的动态变化。在无需标注的情况下，我们推断出随时间演化的类别原型，这些原型充当动态分类头。此外，我们的方法非常轻量，仅修改模型的最后一个线性层。在真实世界分布偏移和合成损坏的实验上，我们证明我们的方法与需要反向传播和访问模型主干的方法相比具有竞争力。我们的模型尤其在小测试批量的情况下表现出色——这是最具挑战性的设置。