一个用于时间序列异常检测的基于预测的综合框架:在Numenta异常基准(NAB)上进行基准测试
A Comprehensive Forecasting-Based Framework for Time Series Anomaly Detection: Benchmarking on the Numenta Anomaly Benchmark (NAB)
AI总结:
本文提出一个统一的基于预测的时间序列异常检测综合框架,整合经典与深度学习方法,并在NAB基准上完成首次全面评估,通过大量实验对比证明LSTM在复杂模式上表现最佳,Informer在效率上更具优势,为未来研究建立了基准。
摘要:
时间序列异常检测对于现代数字基础设施至关重要,但现有方法缺乏系统的跨领域评估。我们提出了一个综合的基于预测的框架,在通用的基于残差的检测接口下,将经典方法(Holt-Winters、SARIMA)与深度学习架构(LSTM、Informer)统一起来。我们的模块化流水线集成了预处理(归一化、STL分解)、四种预测模型、四种检测方法,以及通过预测指标(MAE、RMSE、PCC)和检测指标(Precision、Recall、F1、AUC)的双重评估。我们在Numenta异常基准(58个数据集,7个类别)上进行了首次完整评估,进行了232次模型训练运行和464次检测评估,成功率达到100%。LSTM取得了最佳性能(F1:0.688,在81%的数据集上排名第一或第二),在复杂模式上具有出色的相关性(PCC:0.999)。Informer提供了有竞争力的准确率(F1:0.683),训练速度提高了30%。经典方法在简单的合成数据上以60%的更低成本实现了完美预测,但在真实世界数据集上的F1分数低2-3倍。预测质量主导检测性能:检测方法之间的差异(F1:0.621-0.688)小于预测模型之间的差异(F1:0.344-0.688)。我们的发现提供了基于证据的指导:对于复杂模式使用LSTM,对于效率至关重要的部署使用Informer,对于带有资源限制的简单周期性数据使用经典方法。完整的实现和结果为未来基于预测的异常检测研究建立了基准。
英文摘要:
Time series anomaly detection is critical for modern digital infrastructures, yet existing methods lack systematic cross-domain evaluation. We present a comprehensive forecasting-based framework unifying classical methods (Holt-Winters, SARIMA) with deep learning architectures (LSTM, Informer) under a common residual-based detection interface. Our modular pipeline integrates preprocessing (normalization, STL decomposition), four forecasting models, four detection methods, and dual evaluation through forecasting metrics (MAE, RMSE, PCC) and detection metrics (Precision, Recall, F1, AUC). We conduct the first complete evaluation on the Numenta Anomaly Benchmark (58 datasets, 7 categories) with 232 model training runs and 464 detection evaluations achieving 100\% success rate. LSTM achieves best performance (F1: 0.688, ranking first or second on 81\% of datasets) with exceptional correlation on complex patterns (PCC: 0.999). Informer provides competitive accuracy (F1: 0.683) with 30\% faster training. Classical methods achieve perfect predictions on simple synthetic data with 60 lower cost but show 2-3 worse F1-scores on real-world datasets. Forecasting quality dominates detection performance: differences between detection methods (F1: 0.621-0.688) are smaller than between forecasting models (F1: 0.344-0.688). Our findings provide evidence-based guidance: use LSTM for complex patterns, Informer for efficiency-critical deployments, and classical methods for simple periodic data with resource constraints. The complete implementation and results establish baselines for future forecasting-based anomaly detection research.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。