Awesome Time Series Segmentation Papers

Stars

Star History

Description

This repository contains a reading list of papers on Time Series Segmentation. This repository is still being continuously improved.

As a crucial time series preprocessing technique, semantic segmentation divides poorly understood time series into several discrete and homogeneous segments. This approach aims to uncover latent temporal evolution patterns, detect unexpected regularities and regimes, thereby rendering the analysis of massive time series data more manageable.

Time series segmentation often intertwines with research in many domains. Firstly, the relationship between time series segmentation, time series change point detection, and some aspects of time series anomaly/outlier detection is somewhat ambiguous. Therefore, this repository includes a selection of papers from these areas. Secondly, time series segmentation can be regarded as a process of information compression in time series, hence papers in this field often incorporate concepts from information theory (e.g., using minimum description length to guide the design of unsupervised time series segmentation models). Additionally, the task of decomposing human actions into a series of plausible motion primitives can be addressed through methods for segmenting sensor time series. Consequently, papers related to motion capture from the fields of computer vision and ubiquitous computing are also included in this collection.

Generally, the subjects of unsupervised semantic segmentation can be categorized into:

univariate time series: $1\times T$ , where $T$ is the length of the time series.
multivariate time series: $M\times T$ , where $M$ is the number of variables (channels).
tensor: $N_{1} \times \cdots \times N_{k} \times M\times T$ , where $N_{1} \times \cdots \times N_{k}$ denotes the dimensions other than time and variables.

In the field of time series research, unlike time series forecasting, anomaly detection, and classification/clustering, the number of papers on time series segmentation has been somewhat lukewarm in recent years (this observation may carry a degree of subjectivity from the author). Additionally, deep learning methods do not seem to dominate this area as they do in others. Some classic but solid algorithms remain highly competitive even today, with quite a few originating from the same research group. Therefore, in the following paper list, I will introduce them indexed by well-known researchers and research groups in this field.

Some Additional Information

🚩 2024/4/28: In fact, manually annotating segment points (change points) in large time series datasets is extremely labor-intensive and somewhat subjective. Therefore, the field of time series segmentation lacks large public datasets with ground truth, making it difficult for supervised methods to find sources of training data. Unsupervised time series segmentation also acts to some extent as an automatic annotator of segmentation points, making it easier to implement. Currently, 95% of the research work included in this repository is unsupervised.

🚩 2024/1/27: I have marked some recommended papers / datasets / implementations with 🌟 (Just my personal preference 😉).

Survey & Evaluation

NOTE: the ranking has no particular order.

TYPE	Venue	Paper Title and Paper Interpretation	Code
Dataset	DARLI-AP@EDBT/ICDT '23	Time Series Segmentation Applied to a New Data Set for Mobile Sensing of Human Activities 🌟	MOSAD
Dataset	ECML-PKDD Workshop '23	Human Activity Segmentation Challenge@ECML/PKDD’23 🌟	Challenge Link
Visualization	IEEE TVCG '21	MultiSegVA Using Visual Analytics to Segment Biologging Time Series on Multiple Scales	None
Survey	IEEE J. Sel. Areas Commun. '21	Sequential (Quickest) Change Detection Classical Results and New Directions	None
Survey	Signal Process. '20	Selective review of offline change point detection methods 🌟	Ruptures
Evaluation	Arxiv '20	An Evaluation of Change Point Detection Algorithms 🌟	TCPDBench
Survey	Knowl. Inf. Syst. '17	A survey of methods for time series change point detection 🌟	None
Evaluation	Inf. Syst. '17	An evaluation of combinations of lossy compression and change-detection approaches for time-series data	None
Survey	IEEE Trans Hum. Mach. Syst. '16	Movement Primitive Segmentation for Human Motion Modeling A Framework for Analysis 🌟	None
Survey	EAAI '11	A review on time series data mining	None
Survey	CSUR '11	Time-series data mining	None
Dataset	GI '04	Segmenting Motion Capture Data into Distinct Behaviors 🌟	Website

David Hallac (Stanford)

TYPE	Venue	Paper Title and Paper Interpretation	Code
	KDD Workshop MiLeTS '20	Driver2vec Driver Identification from Automotive Data	Driver2vec
	Adv. Data Anal. Classif. '19	Greedy Gaussian segmentation of multivariate time series 🌟	GGS
	Arxiv '18	MASA: Motif-Aware State Assignment in Noisy Time Series Data	MASA
Ph.D. Thesis	ProQuest '18	Inferring Structure from Multivariate Time Series Sensor Data	None
	KDD '17	Toeplitz Inverse Covariance-Based Clustering of Multivariate Time Series Data 🌟	TICC
	KDD '17	Network Inference via the Time-Varying Graphical Lasso 🌟	TVGL

Shaghayegh Gharghabi (from Eamonn Keogh's Lab, UC Riverside)

TYPE	Venue	Paper Title and Paper Interpretation	Code
	DMKD '19	Domain agnostic online semantic segmentation for multi-dimensional time series 🌟	Floss & datasets)
	ICDM '17	Matrix Profile VIII Domain Agnostic Online Semantic Segmentation at Superhuman Performance Levels 🌟	Floss

Yasuko Matsubara & Yasushi Sakurai (from Sakurai & Matsubara Lab)

Venue	Paper Title and Paper Interpretation	Code
WWW '24	Dynamic Multi-Network Mining of Tensor Time Series 🌟	DMM
WWW '23	Fast and Multi-aspect Mining of Complex Time-stamped Event Streams 🌟	CubeScope
KDD '22	Fast Mining and Forecasting of Co-evolving Epidemiological Data Streams 🌟	None
CIKM '22	Modeling Dynamic Interactions over Tensor Streams	Dismo
CIKM '22	Mining Reaction and Diffusion Dynamics in Social Activities 🌟	None
NeurIPS '21	SSMF Shifting Seasonal Matrix Factorization	ssmf
KDD '20	Non-Linear Mining of Social Activities in Tensor Streams 🌟	None
ICDM '19	Multi-aspect mining of complex sensor sequences 🌟	CubeMarker
KDD '19	Dynamic Modeling and Forecasting of Time-evolving Data Streams	OrbitMap
CIKM '19	Automatic Sequential Pattern Mining in Data Streams	None
KDD '16	Regime Shifts in Streams: Real-time Forecasting of Co-evolving Time Sequences	RegimeCast
WWW '16	Non-linear mining of competing local activities	CompCube
WWW '15	The web as a jungle: Non-linear dynamical systems for co-evolving online activities 🌟	Ecoweb & dataset
SIGMOD '14	AutoPlait Automatic Mining of Co-evolving Time Sequences 🌟	AutoPlait
ICDM '14	Fast and Exact Monitoring of Co-evolving Data Streams	None
KDD '14	FUNNEL Automatic Mining of Spatially Coevolving Epidemics	Funnel

Bryan Hooi (NUS)

TYPE	Venue	Paper Title and Paper Interpretation	Code
	TKDE '22	Time Series Anomaly Detection with Adversarial Reconstruction Networks 🌟	BeatGAN
	IJCAI '19	BeatGAN Anomalous Rhythm Detection using Adversarially Generated Time Series 🌟	BeatGAN
Ph.D. Thesis	ProQuest '19	Anomaly Detection in Graphs and Time Series Algorithms and Applications	None
![multivariate time series