This repository contains the code used to produce the results presented in the IJCNN 2017 paper "DropIn: Making Reservoir Computing Neural Networks Robust to Missing Inputs by Dropout" by D. Bacciu, F. Crecchi (University of Pisa) and D. Morelli (Biobeats LTD).

The paper presents a novel, principled approach to train recurrent neural networks from the Reservoir Computing family that are robust to missing part of the input features at prediction time exploiting Dropout regularization technique to simulate missing inputs during training.

This research software is provided as is and it is not intended to be a comprehensive library for missing data handling in neural networks. The library includes both scripts to reproduce the experiments in the the papaer, as well as well identified, stand-alone object code implementing the DropIN-ESN model.

If you happen to use or modify this code, please remember to cite: D. Bacciu, F. Crecchi, D. Morelli, DropIn: Making Reservoir Computing Neural Networks Robust to Missing Inputs by Dropout, Proceedings of the 30th International Joint Conference on Neural Networks (IJCNN'17), IEEE, 2017.