We are happy to announce that the paper "ReLiNet: Stable and Explainable Multistep Prediction with Recurrent Linear Parameter Varying Networks" has been accepted to the prestigious International Joint Conference on Artificial Intelligence (IJCAI-23).
This paper proposes a novel approach, ReLiNet, combining switched linear systems with recurrent neural networks. ReLiNet achieves multistep prediction accuracy close to the state-of-the-art while retaining the explainability and system-theoretic properties of linear systems. Furthermore, it is demonstrated that the method can be easily extended to guarantee exponential stability for a small tradeoff in prediction accuracy.
A. Baier, D. Aspandi, and S. Staab, "ReLiNet: Stable and Explainable Multistep Prediction with Recurrent Linear Parameter Varying Networks", Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI-23), 2023.