Online sequential extreme learning machine with kernels for nonstationary time series prediction

In this paper, an online sequential extreme learning machine with kernels (OS-ELMK) has been proposed for nonstationary time series prediction. An online sequential learning algorithm, which can learn samples one-by-one or chunk-by-chunk, is developed for extreme learning machine with kernels. A limited memory prediction strategy based on the proposed OS-ELMK is designed to model the nonstationary time series. Performance comparisons of OS-ELMK with other existing algorithms are presented using artificial and real life nonstationary time series data. The results show that the proposed OS-ELMK produces similar or better accuracies with at least an order-of-magnitude reduction in the learning time.