LOCUS - LOcalization with Channel Uncertainty and Sporadic Energy

Step 01: Download this repository

cd LOCUS

Step 02: Create virtual environment

apt-get install python3-venv
python3 -m venv venv
source venv/bin/activate

Step 03: Install requirements

pip install -r requirements.txt

Step 04: Download the data form here. Unzip the downladed file into the <data> folder.

mkdir data
cd data
unzip <downloaded_file>.zip

Note: A small subset of data is provided with the repository to test the pipeline. To train the whole model with the full dataset, please follow the link given above.

Step 05: Create the full-rank feature, $F$

python feature.py --full_rank

Step 06: Train Baseline SELDNet

pyhton train.py

Step 07: Create low-rank feature set, $\tilde{F}$

• Update feature_params from parameters/parameters.py . For example, to create $\tilde{F}$ with $75\%$ of dopping feature_params should be like this:

  feature_params = {
      "is_doping": True,
      "doping_pct": .75,
      "no_dopped_channel": 3
      }

• Run the following:

  python feature.py
  

Step 08: Train LOCUS:

python -W ignore train_locus.py