This repository provides the implementation of the Deep Contrastive One-Class Time Series Anomaly Detection method, called COCA bellow.
The implementation uses the Merlion and the Tsaug libraries.
With the accumulation of large volume time-series data, discerning novel or anomalous events, i.e., outlier, is becoming more and more important. With the scarcity of labels, anomaly detection from time-series data with temporal dynamics is a very challenging task. As a kind of unsupervised learning, contrastive self-supervised learning has been made significant progress in a variety of applications, and applying it to unlabeled time series data is promising but still open. Existing approaches are usually based on a single assumption of normality or require pre-training; they do not fully consider properties of overall normality and are limited by the quality of the representations. This paper proposes a deep Contrastive One-Class Anomaly detection of time series (COCA), combining three normality assumptions. First, to make it easier to distinguish anomalies from normal samples, the original time series data are expanded via data augmentation and then transformed into two correlated views. Second, to learn temporal dependencies which are key characteristics of time series, a powerful Seq2Seq model is used in latent space to reconstruct representations of each time step. Last, we propose a contrastive one-class loss function to build a classifier with only one stage, from which an anomaly score is defined. Extensive experiments conducted on four real-world time-series datasets show the superior performance of the proposed methods over the state-of-the-arts. The code is publicly available at https://github.com/ruiking04/COCA.
This code is based on Python 3.8, all requires are written in requirements.txt. Additionally, we should install saleforce-merlion and ts_dataset as Merlion suggested.
git clone https://github.com/salesforce/Merlion.git
cd Merlion
pip install salesforce-merlion
pip install -e Merlion/ts_datasets/
pip install -r requirements.txt
This directory contains experiment parameters for all models on NAB, IOpsCompetition, UCR, SMAP datasets.
Source code of OCSVM, DeepSVDD, CPC, TS-TCC and COCA（OC_CL） models.
Directory where the experiment results and checkpoint are saved.
This implementation is based on Deep-SVDD-PyTorch, Contrastive-Predictive-Coding-PyTorch and TS-TCC