We proposed a new forecasting model using self-organizing fuzzy inference network (SOFIN) to solve the problem for forecasting Mackey-Glass chaotic time series, which generates the network structure and parameters automatically by means of fuzzy c-means clustering method (FCM) and learning algorithm from examples (LAE).
First, the FCM is used to obtain the centers of activation functions in the first layer of network.
Second, the LAE is used to obtain the weights connecting the second layer with the third layer of the network which represents the fuzzy rules. After the network structure is formed, the output layer of the network is trained by using the gradient descent method.
Finally, the proposed forecasting model is a time series forecasting model with more forecasting accuracy than the other typical forecasting techniques through the forecasting analysis of Mackey-Glass chaotic time series.
The result was published in "Journal of the institution of Engineer(India):series B" under the title of "Mackey-Glass Chaotic Time Series Forecasting by using Self-Organizing Fuzzy Inference Network"(https://doi.org/10.1007/s40031-023-00855-6).
