Clej / FedForecast

Contain a notebook and python function to perform federated averaging on foreecasting models (e.g LSTM).

Application of Federated Averging algorithm to train LSTM models for time series forecasting. Notebook is written in french.

Libraries

Libraries used in this repo:

sklearn: 0.24.2 pytorch: 1.9.0 numpy: 1.19.5 matplotlib: 3.2.2

Datasets

See datasets directory to get the datasets used. In the scripts and notebook, a dataset is a multivariate time-series stored in a numpy array with size (T_size, n_variables), where T_size is the number of time steps (same along the time series components) and n_variables is the dimension of a single temporal sample.

Author

Clément Lejeune.

References

Datasets from: H. F. Yu, N. Rao, and I. S. Dhillon, “Temporal regularized matrix factorization for high-dimensional time series prediction,” in NIPS, 2016, pp. 847–855.

Federated Averaging algorithm: H. B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. Agüera y Arcas, “Communication-Efficient Learning of Deep Networks from Decentralized Data,” in AISTATS, 2017, vol. 54.