This paper is dedicated to the long-term, or multi-step-ahead, time series prediction problem. We propose a novel method for training feed-forward neural networks, such as multilayer perceptrons, with tapped delay lines. Special batch calculation of derivatives called Forecasted Propagation Through Time and batch modification of the Extended Kalman Filter are introduced. Experiments were carrie...

The paper describes the identification of nonlinear dynamic systems with a Gaussian process prior model. This approach is an example of a probabilistic, non-parametric modelling. Gaussian process model can be considered as the special case of radial basis function network and as such an alternative to neural networks or fuzzy black box models. An attractive feature of Gaussian process model is ...

Parametric multiple model techniques have recently been proposed for the modelling of non–linear systems and use in nonlinear control. Research effort has focused on issues such as the selection of the structure, constructive learning techniques, computational issues, the curse of dimensionality, off–equilibrium behavior etc. To reduce these problems, the use of non–parametrical modelling appro...

This paper evaluates the predictability of monthly stock return using out-of-sample (multi-step ahead and dynamic) prediction intervals. Past studies have exclusively used point forecasts, which are of limited value since they carry no information about the intrinsic predictive uncertainty associated. We compare empirical performances of alternative prediction intervals for stock return generat...

Activated sludge process has been widely adopted to remove pollutants in wastewater treatment plants (WWTPs). However, stable operation of activated sludge process is often compromised by the occurrence of filamentous bulking. The aim of this study is to build a proper model for timely diagnosis and prediction of filamentous sludge bulking in an activated sludge process. This study developed a ...

Multi-task learning employs shared representation of knowledge for learning multiple instances from the same or related problems. Time series prediction consists of several instances that are defined by the way they are broken down into fixed windows known as embedding dimension. Finding the optimal values for embedding dimension is a computationally intensive task. Therefore, we introduce a ne...

A multi-step ahead predictive filter for missing data handling is presented in this paper. This is a simple FIR filter, useful for time and memory critical applications. Furthermore, our proposed algorithm is formulated in such a way that only one set of FIR weights are tuned (by steepest gradient descent method), memorised and used by the system for different numbers of steps ahead in predicti...

The task of forecasting a time series over a long horizon is commonly tackled b y iterating one-step-ahead predictors.Despite the popularity that this approach gained in the prediction communit y, its design is still plagued by a number of important unresolved issues, the most important being the accumulation of prediction errors. We introduce a local method to learn one-step-ahead predictors w...