I-SEA: Importance Sampling and Expected Alignment-based Deep Distance Metric Learning for Time Series Analysis and Embedding
The codebase is written in python, C, and MATLAB.
The MATLAB-based baselines are under TimeSeriesMetricLearning/matlab
.
The python (learning-based) baselines are under TimeSeriesMetricLearning/python_pytorch
, and the following are the commands to compile the C-based libraries for I-SEA/DECADE.
- DECADE
gcc -c -fPIC edist_c_path_py.c -o edist_c_path_py.o
gcc edist_c_path_py.o -shared -o edist_c_path.so
- DECADE IS
gcc -c -fPIC edist_is_c_path_py.c -o edist_is_c_path_py.o
gcc edist_is_c_path_py.o -shared -o edist_is_c_path.so
gcc -c -fPIC -I/usr/include edist_is_c_path_py.c -o edist_is_c_path_py.o
gcc edist_is_c_path_py.o -shared -o edist_is_c_path.so -L/usr/lib/x86_64-linux-gnu -lgsl -lgslcblas -lm
- DECADE IS (KDE)
gcc -c -fPIC -I/usr/include kde-edist_is_c_path_py.c -o kde-edist_is_c_path_py.o
gcc kde-edist_is_c_path_py.o -shared -o kde-edist_is_c_path_py.so -L/usr/lib/x86_64-linux-gnu -lgsl -lgslcblas -lm