Time Series Subsequence Similarity Search Under Dynamic Time Warping Distance on the Intel Many-core Accelerators

Author(s):  
Aleksandr Movchan ◽  
Mikhail Zymbler
IEEE Access ◽  
2019 ◽  
Vol 7 ◽  
pp. 163644-163653 ◽  
Author(s):  
Zhengxin Li ◽  
Jiansheng Guo ◽  
Hailin Li ◽  
Tao Wu ◽  
Sheng Mao ◽  
...  

Author(s):  
Я.А. Краева ◽  
М.Л. Цымблер

В настоящее время поиск похожих подпоследовательностей требуется в широком спектре приложений интеллектуального анализа временных рядов: моделирование климата, финансовые прогнозы, медицинские исследования и др. В большинстве указанных приложений при поиске используется мера схожести Dynamic Time Warping (DTW), поскольку на сегодняшний день научное сообщество признает меру DTW одной из лучших для большинства предметных областей. Мера DTW имеет квадратичную вычислительную сложность относительно длины искомой подпоследовательности, в силу чего разработан ряд параллельных алгоритмов ее вычисления на устройствах FPGA и многоядерных ускорителях с архитектурами GPU и Intel MIC. В настоящей статье предлагается новый параллельный алгоритм для поиска похожих подпоследовательностей в сверхбольших временных рядах на кластерных системах с узлами на базе многоядерных процессоров Intel Xeon Phi поколения Knights Landing (KNL). Вычисления распараллеливаются на двух уровнях: на уровне всех узлов кластера - с помощью технологии MPI и в рамках одного узла кластера - с помощью технологии OpenMP. Алгоритм предполагает использование дополнительных структур данных и избыточных вычислений, позволяющих эффективно задействовать возможности векторизации вычислений на процессорных системах Phi KNL. Эксперименты, проведенные на синтетических и реальных наборах данных, показали хорошую масштабируемость алгоритма. Nowadays, the subsequence similarity search is required in a wide range of time series mining applications: climate modeling, financial forecasts, medical research, etc. In most of these applications, the Dynamic Time Warping (DTW) similarity measure is used, since DTW is empirically confirmed as one of the best similarity measures for the majority of subject domains. Since the DTW measure has a quadratic computational complexity with respect to the length of query subsequence, a number of parallel algorithms for various many-core architectures are developed, namely FPGA, GPU, and Intel MIC. In this paper we propose a new parallel algorithm for subsequence similarity search in very large time series on computer cluster systems with nodes based on Intel Xeon Phi Knights Landing (KNL) many-core processors. Computations are parallelized on two levels as follows: by MPI at the level of all cluster nodes and by OpenMP within a single cluster node. The algorithm involves additional data structures and redundant computations, which make it possible to efficiently use the capabilities of vector computations on Phi KNL. Experimental evaluation of the algorithm on real-world and synthetic datasets shows that the proposed algorithm is highly scalable.


Author(s):  
Aleksandra Rutkowska ◽  
Magdalena Szyszko

AbstractThis study provides an application of dynamic time warping algorithm with a new window constraint to assess consumer expectations’ information content regarding current and future inflation. Our study’s contribution is the novel application of DTW for testing expectations’ forward-lookingness. Additionally, we modify the algorithm to adjust it for a specific question on the information content of our data. The DTW overcomes constraints of the standard tool that examines forward-lookingness: DTW does not impose assumptions on time series properties. In empirical study we cover seven European counties and compare the DTW outcomes with the results of previous studies in these economies using a standard methodology. The research period covers 2001 to mid-2018. Application of DTW provides information on the degree of expectations’ forward-lookingness. The result, after standardization, are similar to the standard parameters of hybrid specification of expectations. Moreover, the rankings of most forward-looking consumers are replicated. Our results confirm the economic intuition, and they do not contradict previous studies.


Energies ◽  
2021 ◽  
Vol 14 (13) ◽  
pp. 4024
Author(s):  
Krzysztof Dmytrów ◽  
Joanna Landmesser ◽  
Beata Bieszk-Stolorz

The main objective of the study is to assess the similarity between the time series of energy commodity prices and the time series of daily COVID-19 cases. The COVID-19 pandemic affects all aspects of the global economy. Although this impact is multifaceted, we assess the connections between the number of COVID-19 cases and the energy commodities sector. We analyse these connections by using the Dynamic Time Warping (DTW) method. On this basis, we calculate the similarity measure—the DTW distance between the time series—and use it to group the energy commodities according to their price change. Our analysis also includes finding the time shifts between daily COVID-19 cases and commodity prices in subperiods according to the chronology of the COVID-19 pandemic. Our findings are that commodities such as ULSD, heating oil, crude oil, and gasoline are weakly associated with COVID-19. On the other hand, natural gas, palm oil, CO2 allowances, and ethanol are strongly associated with the development of the pandemic.


Sign in / Sign up

Export Citation Format

Share Document