Deploying pretrained visual models in real-world environments often suffers from significant performance degradation due to the diversity of testing scenarios. Continuous adaptation of learning models on edge devices via unlabeled data collected from the target domain is highly effective for boosting generalization capability. However, gradient-backpropagation-based optimization of the massive parameters in deep neural networks is vastly more time-consuming than forward inference, rendering online learning infeasible on low-power edge devices. To address this critical challenge, we propose a lightweight gradient-free forward-memorizing framework, namely MemFlow, which leverages a frozen backbone and enables efficient fine-tuning of the mapping between features and predictions. Specifically, MemFlow employs randomly connected neurons to memorize feature-label associations; within the network, spiking signals are propagated, and predictions are generated by associating neuron-stored memories according to their confidence levels. More notably, MemFlow supports reinforced memorization of feature mappings using unlabeled data, thereby enabling rapid adaptation to new domains. Extensive experiments on four real-world cross-domain datasets demonstrate that MemFlow achieves performance improvements of up to 10\% while consuming less than 1\% of the computational time required by traditional domain adaptation methods.The code is available at https://github.com/so-link/MemFlow.

翻译：在真实环境中部署预训练视觉模型时，因测试场景的多样性常导致显著性能下降。通过从目标域收集的无标签数据在边缘设备上持续自适应学习模型，是提升泛化能力的高效手段。然而，基于梯度反向传播的深度神经网络海量参数优化耗时远超前向推理，导致在线学习在低功耗边缘设备上难以实现。为应对这一关键挑战，我们提出轻量级无梯度前向记忆框架MemFlow，该框架利用冻结主干网络实现特征与预测间映射的高效微调。具体而言，MemFlow采用随机连接神经元存储特征-标签关联，网络内传播脉冲信号，并根据置信度关联神经元存储的记忆生成预测。更值得注意的是，MemFlow支持利用无标签数据增强特征映射的记忆强化，从而快速适应新领域。在四个真实跨域数据集上的实验表明，MemFlow在传统域自适应方法所需时间不到1%的计算开销下，实现了最高10%的性能提升。代码开源地址：https://github.com/so-link/MemFlow。