This paper studies a connection between RNNs and kenel-based learning. This leads to development of new algorithms for sequential learning and results are validated experimentally. Overall, this represents an interesting contribution for Neurips conference.