lemire / lbimproved

Dynamic Time Warping (DTW) library implementing lower bounds (LB_Keogh, LB_Improved...)

LBImproved C++ Library

This library comes in the form of one short C++ header file. The documentation is in the C++ comments and in this file.

Key feature

1. Fast Dynamic Time Warping nearest neighbor retrieval.

2. Implementations of LB Koegh and LB Improved

3. Companion to the following paper :

Daniel Lemire, Faster Retrieval with a Two-Pass Dynamic-Time-Warping Lower Bound, Pattern Recognition 42 (9), pages 2169-2180, 2009. http://arxiv.org/abs/0811.3301

Comments about this paper by Keogh's team:

 To our knowledge, there is only one paper that
 offers a plausible speedup based on a tighter 
 lower bound—Lemire (2009) suggests a mean speedup 
 of about 1.4 based on a tighter bound. 
 These results are reproducible, and testing on 
 more general data sets we obtained similar 
 results (...) (Wang et al. 2013)

BUILD

type "make"

make
./unittesting
./benchmark
./example

Simple code example

See example.cpp.

Other libraries

• dtwclust is an R Package for Time Series Clustering Along with Optimizations for DTW