LAMP: Learned Approximate Matrix Profile

This code provides an implementation of LAMP training and inference as described in "Time Series Mining in the Face of Fast Moving Streams using a Learned Approximate Matrix Profile"

Usage:

# For training + inference
python train_neural_net_LAMP.py <matrix profile window size> <input data path> <logging output path> <pretrained weights file (optional)> <initial epoch (optional)>
# For inference only
python train_neural_net_LAMP.py <matrix profile window size> <input data path> <logging output path> <pretrained weights file>

Data Format

Input data must be organized into a mat file with a specific set of variables defined: \ ts_train (time series input for training)
mp_train (matrix profile output for training)
ts_val (validation time series)
mp_val (validation matrix profile)
ts_test (testing time series)
mp_test (testing matrix profile) \