A paper written by Dr. Stefan Heinrich has been accepted in the 2020 IEEE International Conference on Systems, Man, and Cybernetics. It proposes a novel mechanism to optimize time scales of a recurrent neural network through learning.
- Stefan Heinrich and Yukie Nagai, “Learning Timescales in Gated and Adaptive Continuous Time Recurrent Neural Networks,” in Proceedings of the 2020 IEEE International Conference on Systems, Man, and Cybernetics, accepted, 2020.