Fed-MEx is the Federated Version of the MEx Human Activity Recognition dataset. Pressure mat subset is used in this dataset.

MEx is a publicly available exercise recognition dataset collected with 30 subjects performing 7 different physiotherapy exercises. The MEx dataset has 934 data samples from the pressure mat subset of the MEx dataset. Each client has a random amount of samples for only 2 exercise classes. A pressure mat data sample contains a sequence of heat maps (size 5 x 16 x 16) recorded for 5 seconds with 1Hz frequency. MEx has previously been used for personalised activity recognition research(1) and forms an interesting contrast to the other image and text datasets.

To generate the dataset use the pm folder from the downloaded MEx dataset and copy it to the root directory.

This dataset is compatible with the FedProx, FedSim implemenations. If you are using the same experiment setup simply add the following into the main.py,

DATASETS = [....., 'mex']

MODEL_PARAMS = {
    ..... , 
    'mex.mclr': (7,), 
    'mex.cnn': (7,), 
}

Reference models are availble in the FedSim implementation here.

1 - WIJEKOON, A., WIRATUNGA, N., COOPER, K. and BACH, K. 2020. Learning to recognise exercises for the self-management of low back pain. In Barták, R. and Bell, E. (eds.). Proceedings of the 33rd International Florida Artificial Intelligence Research Society (FLAIRS) 2020 conference (FLAIRS-33), 17-20 May 2020, Miami Beach, USA. Palo Alto: AAAI Press [online], pages 347-352.

UCI ML Repository

GitHub

Fed-Goodreads