Abstract:Time series similarity search is an essential operation in time series data mining and has received much higher interest along with the growing popularity of time series data. Although many algorithms to solve this problem have been investigated, there is a challenging demand for supporting similarity search in a fast and accurate way. In this paper, we present a novel approach, TS2BC, to perform time series similarity search efficiently and effectively. TS2BC uses binary code to represent time series and measures the similarity under the Hamming Distance. Our method is able to represent original data compactly and can handle shifted time series and work with time series of different lengths. Moreover, it can be performed with reasonably low complexity due to the efficiency of calculating the Hamming Distance. We extensively compare TS2BC with state-of-the-art algorithms in classification framework using 61 online datasets. Experimental results show that TS2BC achieves better or comparative performance than other the state-of-the-art in accuracy and is much faster than most existing algorithms. Furthermore, we propose an approximate version of TS2BC to speed up the query procedure and test its efficiency by experiment.

Binary Sequences With Good Aperiodic Autocorrelations Using Cross-Entropy Method

An Efficient Heuristics Search For Binary Sequences With Good Aperiodic Autocorrelations

An efficient method for time series similarity search using binary code representation and hamming distance

Efficient Heuristic Method of Search for Binary Sequences with Good Aperiodic Autocorrelations

A General Construction of Binary Sequences with Optimal Autocorrelation

New Binary Sequences with Good Correlation Based on High-Order Difference and Interleaving Techniques

Some New Results on the Cross Correlation of &Lt;inline-Formula> &Lt;tex-Math Notation="tex">$m$ &Lt;/tex-Math></inline-formula>-sequences

Sequences with small correlation

On Detecting Feasible Periodicity for Periodic Event in Binary Data Series.

The Cross-Entropy Method and Its Application to Inverse Problems

A class of balanced binary sequences with two-valued non-zero autocorrelation sum and good crosscorrelation sum

A New Class of Ternary and Quaternary Sequences with Two-Level Autocorrelation

Some Notes on Constructions of Binary Sequences with Optimal Autocorrelation

Efficient Generation of Low Autocorrelation Binary Sequences

An Improved Multi-Objective Cross-Entropy Method For Inverse Problems

Ternary Perfect Sequences with Three-Valued Cross-Correlation.

A Tutorial on the Cross-Entropy Method

Aperiodic Array Synthesis Using the Cross Entropy Method

New Optimal Binary Sequences with Period $4p$ via Interleaving Ding-Helleseth-Lam Sequences

Computational Searching of Long Skew-symmetric Binary Sequences with High Merit Factors