面向时间序列异常检测的基础自编码器
Towards Foundation Auto-Encoders for Time-Series Anomaly Detection
AI总结：
本文提出基于VAE和膨胀卷积的基础自编码器FAE，通过在海量时间序列上预训练实现通用建模与零样本异常检测，并在多个数据集上验证了其初步效果。
摘要：
受大规模预训练基础模型成功的启发，我们研究了一种新颖的时间序列建模方法。我们提出了FAE（Foundation Auto-Encoders，基础自编码器），这是一种基于变分自编码器（VAEs）的基础生成式AI模型，用于时间序列数据中的异常检测。所谓“基础”，是指该模型在海量时间序列数据上进行预训练，能够学习复杂的时间模式，从而有助于对先前未见过的数据集进行准确建模、预测和异常检测。FAE利用VAEs和膨胀卷积神经网络（DCNNs）构建了一个通用的单变量时间序列建模模型，最终有望在开箱即用、零样本异常检测应用中表现良好。我们介绍了FAE的主要概念，并在来自不同领域的多个多维时间序列数据集上展示了初步结果，其中包括一个来自运营中移动互联网服务提供商（ISP）的真实数据集，以及著名的KDD 2021异常检测数据集。
英文摘要：
We investigate a novel approach to time-series modeling, inspired by the successes of large pretrained foundation models. We introduce FAE (Foundation Auto-Encoders), a foundation generative-AI model for anomaly detection in time-series data, based on Variational Auto-Encoders (VAEs). By foundation, we mean a model pretrained on massive amounts of time-series data which can learn complex temporal patterns useful for accurate modeling, forecasting, and detection of anomalies on previously unseen datasets. FAE leverages VAEs and Dilated Convolutional Neural Networks (DCNNs) to build a generic model for univariate time-series modeling, which could eventually perform properly in out-of-the-box, zero-shot anomaly detection applications. We introduce the main concepts of FAE, and present preliminary results in different multi-dimensional time-series datasets from various domains, including a real dataset from an operational mobile ISP, and the well known KDD 2021 Anomaly Detection dataset.
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。
