Time series analysis by similarity in large scale

• Grant number: 13/26151-5
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: July 01, 2014
• End date: September 30, 2017
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Agreement: Coordination of Improvement of Higher Education Personnel (CAPES)
• Principal Investigator: Gustavo Enrique de Almeida Prado Alves Batista
• Grantee: Diego Furtado Silva
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated scholarship(s): 15/07628-0 - Speeding up All-Pairwise Dynamic Time Warping distance matrix for time series mining, BE.EP.DR

Abstract

Time series are ubiquitous in day-by-day of human beings. Consequently, the area of time series analysis has attracted the attention and effort of many researchers around the world. Due to the great interest in time series, many analysis methods have been proposed to in recent decades. Several of these methods have one thing in common: in their cores, there is a similarity function used as the main way to compare the time series. Several studies in the area show that the distance measure Dynamic Time Warping (DTW) is one of the most appropriate measures to compare time series in a wide range of application domains. However, due to its complexity runtime, the application of this measure goes against one of the greatest current challenges in the area: the analysis of large temporal databases. In this project, we are interested in developing techniques that can be used in various tasks of large-scale analysis of time series using similarity. For this purpose, the main hypothesis of this work is that the efficiency of the algorithm that calculates the DTW distance measure can be improved taking into account only the two series to be compared. Such improvements of the DTW algorithm should be able to provide a significantly better time performance in different tasks of time series analysis. In this project, we describe the objectives and the tasks to be done to achieve them. In addition, we present the basic concepts of time series analysis and its applications in large volumes of data. (AU)

Scientific publications

(6)

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

SILVA, DIEGO F.; BATISTA, GUSTAVO E. A. P. A.; WANI, MA; KANTARDZIC, M; SAYEDMOUCHAWEH, M; GAMA, J; LUGHOFER, E. Elastic Time Series Motifs and Discords. 2018 17TH IEEE INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND APPLICATIONS (ICMLA), v. N/A, p. 6-pg., 2018-01-01. (16/04986-6, 13/26151-5, 17/24340-6)

SILVA, DIEGO F.; GIUSTI, RAFAEL; KEOGH, EAMONN; BATISTA, GUSTAVO E. A. P. A.. Speeding up similarity search under dynamic time warping by pruning unpromising alignments. DATA MINING AND KNOWLEDGE DISCOVERY, v. 32, n. 4, p. 988-1016, JUL 2018. (16/04986-6, 12/08923-8, 13/26151-5)

SILVA, DIEGO F.; SOUZA, VINICIUS M. A.; ELLIS, DANIEL P. W.; KEOGH, EAMONN J.; BATISTA, GUSTAVO E. A. P. A.. Exploring Low Cost Laser Sensors to Identify Flying Insect Species Evaluation of Machine Learning and Signal Processing Methods. JOURNAL OF INTELLIGENT & ROBOTIC SYSTEMS, v. 80, n. 1, SI, p. S313-S330, DEC 2015. (11/17698-5, 12/50714-7, 13/26151-5)

SILVA, DIEGO F.; YEH, CHIN-CHIA M.; ZHU, YAN; BATISTA, GUSTAVO E. A. P. A.; KEOGH, EAMONN. Fast Similarity Matrix Profile for Music Analysis and Exploration. IEEE TRANSACTIONS ON MULTIMEDIA, v. 21, n. 1, p. 29-38, JAN 2019. (16/04986-6, 13/26151-5)

SILVA, DIEGO F.; SOUZA, VINICIUS M. A.; ELLIS, DANIEL P. W.; KEOGH, EAMONN J.; BATISTA, GUSTAVO E. A. P. A.. Exploring Low Cost Laser Sensors to Identify Flying Insect Species Evaluation of Machine Learning and Signal Processing Methods. JOURNAL OF INTELLIGENT & ROBOTIC SYSTEMS, v. 80, p. 18-pg., 2015-12-01. (13/26151-5, 12/50714-7, 11/17698-5)

GIUSTI, RAFAEL; SILVA, DIEGO F.; BATISTA, GUSTAVO E. A. P. A.; IEEE. Improved Time Series Classification with Representation Diversity and SVM. 2016 15TH IEEE INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND APPLICATIONS (ICMLA 2016), v. N/A, p. 6-pg., 2016-01-01. (13/26151-5, 12/08923-8)

Academic Publications

(References retrieved automatically from State of São Paulo Research Institutions)

SILVA, Diego Furtado. Large scale similarity-based time series mining. 2017. Doctoral Thesis - Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB) São Carlos.