发表机构
The Hong Kong University of Science and Technology (GZ); Central South University; The Hong Kong University of Science and Technology（香港科技大学（广州）; 中南大学; 香港科技大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对时间序列学习中样本可靠性异质性导致的过/欠正则化问题，提出容量中心调制（CCM）原则及样本自适应容量调制（SACM）框架，利用频谱稀疏性分配逐样本丢弃概率，在301个数据集-骨干网络对上平均降低预测MSE 6.7%，提升分类准确率3.04%和F1 17.05%，零测试开销。
AI 中文摘要
样本级可靠性异质性在深度时间序列学习中普遍存在。标准训练流程对所有样本应用统一的正则化设置，这可能导致对损坏样本正则化不足，而对干净样本过度约束。常见的鲁棒性方法在数据空间中过滤观测值或对潜在表示施加先验。我们提出容量中心调制（CCM）作为一种互补的、样本自适应的正则化原则。在此原则下，我们引入SACM（样本自适应容量调制），一个任务无关的框架，利用频谱稀疏性沿内部激活路径分配逐样本的丢弃概率。SACM无需架构重新设计即可集成到现有骨干网络中，并保留确定性推理流程。在覆盖9个预测、32个分类和4个异常检测数据集的301个真实世界数据集-骨干网络对上，SACM相对于未修改的骨干网络平均降低预测均方误差6.7%，并分别提高分类准确率和点调整F1分数3.04%和17.05%，且测试时零开销。
英文摘要
Sample-level reliability heterogeneity is common in deep time series learning. Standard training pipelines apply a uniform regularization setting to all samples, which can under-regularize corrupted samples and over-restrict clean samples. Common robustness approaches filter observations in data space or impose priors on latent representations. We propose Capacity-Centric Modulation (CCM) as a complementary, sample-adaptive regularization principle. Under this principle, we introduce SACM (Sample-Adaptive Capacity Modulation), a task-agnostic framework that exploits spectral sparsity to assign sample-wise dropout probabilities along internal activation paths. SACM integrates into existing backbones without architectural redesign and preserves the deterministic inference pipeline. Across 301 real-world dataset-backbone pairs covering 9 forecasting, 32 classification, and 4 anomaly-detection datasets, SACM reduces forecasting MSE by 6.7% on average and improves classification accuracy and point-adjusted F1 by 3.04% and 17.05%, respectively, relative to unmodified backbones, with zero test-time overhead.
