Article
Engineering, Electrical & Electronic
Wei Li, Ruliang He, Binbin Liang, Fan Yang, Songchen Han
Summary: The CDC-DTW algorithm proposed in this study is a method for measuring the similarity of time series with different sampling frequencies. By designing adaptive local context windows and using the technique of local spatial-temporal context density consistency, it overcomes the limitations of conventional DTW algorithms and achieves high accuracy in similarity measure, as demonstrated in extensive experimental results on 128 gold-standard UCR datasets.
IEEE SIGNAL PROCESSING LETTERS
(2023)
Article
Computer Science, Information Systems
Hailin Li
Summary: Dynamic time warping combined with time weight analysis can better reflect the importance of different time points, which is significant for time series data mining.
INFORMATION SCIENCES
(2021)
Article
Computer Science, Artificial Intelligence
Qingzhen Zhang, Chaoqi Zhang, Langfu Cui, Xiaoxuan Han, Yang Jin, Gang Xiang, Yan Shi
Summary: Dynamic time warping (DTW) is one of the most important similarity measurement methods for time series analysis. This paper proposes a time series similarity measurement method based on series decomposition and fast DTW, which calculates the similarity between different components in the time series and amplifies the impact of the more important component to obtain a comprehensive similarity measurement result.
APPLIED INTELLIGENCE
(2023)
Article
Operations Research & Management Science
Pierpaolo D'Urso, Livia De Giovanni, Riccardo Massari
Summary: This paper proposes a robust clustering method for multivariate financial time series, adopting a fuzzy approach and using the Partitioning Around Medoids strategy to consider dynamic time warping distance for neutralizing the negative effects of outliers. The method utilizes a suitable trimming procedure to identify financial time series distant from the data bulk, and is applied to stocks in the FTSE MIB index to identify common time patterns and outliers.
ANNALS OF OPERATIONS RESEARCH
(2021)
Article
Computer Science, Artificial Intelligence
Ahmed Shifaz, Charlotte Pelletier, Francois Petitjean, Geoffrey I. Webb
Summary: This paper presents multivariate versions of seven commonly used elastic similarity and distance measures for time series data analytics. These measures can compensate for misalignments in the time axis of time series data. The paper adapts two existing strategies used in multivariate Dynamic Time Warping to these measures. Demonstrating their utility in multivariate time series classification using the nearest neighbor classifier, the paper shows that each measure achieves the highest accuracy on at least one dataset, supporting the value of developing a suite of multivariate similarity and distance measures. The paper also constructs a nearest neighbor-based ensemble of the measures, which proves to be competitive with other state-of-the-art single-strategy multivariate time series classifiers.
KNOWLEDGE AND INFORMATION SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Chi Zhang, Hadi Fanaee-T, Magne Thoresen
Summary: This study proposes a method for handling irregularly sampled and unequal length electronic health record time series using dynamic time warping and tensor decomposition to learn the latent structure of patient data for patient representation and in-hospital mortality prediction. The research demonstrates outstanding classification performance on two patient cohorts from the MIMIC-III database and provides a detailed analysis of feature importance.
DATA MINING AND KNOWLEDGE DISCOVERY
(2021)
Article
Computer Science, Information Systems
Irati Rasines, Anthony Remazeilles, Miguel Prada, Itziar Cabanes
Summary: This paper proposes an innovative algorithm called MCA-CDTW for averaging multivariate time series with different lengths based on Constrained Dynamic Time Warping (CDTW). The algorithm utilizes CDTW for non-linear alignment and employs Minimum Cost Averaging (MCA) technique for obtaining equal-length time series. Compared to other averaging techniques, MCA-CDTW provides smooth mean curves even with large deviations and reduces algorithm complexity. Experimental results on different databases demonstrate the effectiveness of the proposed algorithm in achieving smooth average trajectories.
Article
Computer Science, Artificial Intelligence
Xiaoyu He, Suixiang Shi, Xiulin Geng, Lingyu Xu
Summary: A novel Information-aware Attention Dynamic Synergy Network (IADSN) is proposed in this study, which effectively addresses the challenges in multivariate time series forecasting through a multi-dimensional attention system and an attention dynamic synergy strategy, achieving higher predictive performance and dynamic trend preservation.
Article
Computer Science, Software Engineering
Oscar Escudero-Arnanz, Antonio G. Marques, Cristina Soguero-Ruiz, Inmaculada Mora-Jimenez, Gregorio Robles
Summary: dtwParallel is a Python package that calculates the DTW distance between a collection of MTS. It incorporates functionalities from current DTW libraries and introduces new features such as parallelization and computation of similarity values. It can handle data with different types of features and is designed for use in education, research, and industry. The package's source code and documentation can be found at https://github.com/oscarescuderoarnanz/dtwParallel.
Article
Computer Science, Artificial Intelligence
Lianpeng Qiu, Cuipeng Qiu, Chengyun Song
Summary: In this paper, a method called ESDTW is proposed for fast and accurate alignment of time series. It introduces local extrema to represent the original time series and aligns the descriptors of extrema shape using DTW. Experimental results show that ESDTW achieves more accurate warping paths compared to other methods, and when combined with the nearest neighbor classifier, it achieves higher classification accuracy.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Computer Science, Artificial Intelligence
Haowen Zhang, Yabo Dong, Duanqing Xu
Summary: Time series classification is a fundamental problem that often requires tuning of numerous parameters, making training time-consuming and adjustment challenging, especially in real-time applications and unseen datasets. This paper introduces a parameter-light algorithm, MDTW, which outperforms other methods by providing faster classification speed and minimal loss in accuracy, while being applicable to previously unseen datasets.
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Abdelmadjid Lahreche, Bachir Boucheham
Summary: The problem of similarity measures within the field of time series classification has led to the development of a new parameter-free measure called LE-DTW, which is designed to quickly and accurately assess similarity between long time series. Experimental results show that LE-DTW performs better than DTW for long time series, while also providing competitive results against popular distance based classifiers. In terms of efficiency, LE-DTW is significantly faster than DTW.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Business
Yanfei Kang, Evangelos Spiliotis, Fotios Petropoulos, Nikolaos Athiniotis, Feng Li, Vassilios Assimakopoulos
Summary: Accurate forecasts are crucial for supporting modern companies in decision-making. This paper introduces a novel data-centric approach called 'forecasting with cross-similarity' to address model uncertainty in a model-free manner. By searching for similar patterns from a reference set, rather than extrapolating, this approach aggregates future paths of similar series to obtain forecasts for the target series. It allows similarity-based forecasting on short series and demonstrates competitive accuracy in real data scenarios.
JOURNAL OF BUSINESS RESEARCH
(2021)
Article
Mathematics, Interdisciplinary Applications
Guangyu Yang, Shuyan Xia
Summary: This paper introduces the recurrence network method and its application in time series analysis. Due to the limitations of irregular sampling, the dynamic time warping method is proposed to calculate distances between time series segments. In order to deal with the sparse sampling area, a weighted dynamic time warping method is further proposed. Two case studies demonstrate the effectiveness of the proposed method.
INTERNATIONAL JOURNAL OF BIFURCATION AND CHAOS
(2023)
Article
Computer Science, Artificial Intelligence
Hailin Li, Tian Du
Summary: This study proposes an MTS clustering method based on a component relationship network (CRN), which consists of a multi-relationship network (MRN) mapping an MTS dataset and utilizing non-negative matrix factorization. By incorporating an improved penalty-coefficient dynamic time-warping algorithm, the method effectively measures the similarity between asynchronous MTS data and improves the accuracy and quality of clustering. Through experiments, it is demonstrated that the proposed method outperforms other clustering methods by considering component correlations and parameter optimization.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
