L1 regularization is better than L2 for learning and predicting chaotic systems

Z. Szabo, A. Lorincz

Emergent behaviors are in the focus of recent research interest. It is then of considerable importance to investigate what optimizations suit the learning and prediction of chaotic…