Traffic prediction using a Generic Self-Evolving Takagi-Sugeno-Kang (GSETSK) fuzzy neural network

This paper analyses traffic prediction based on a Generic Self-Evolving Takagi-Sugeno-Kang (GSETSK) fuzzy neural network. Traffic prediction is a problem that requires online adaptive systems with high accuracy performance. The proposed GSETSK framework can learn incrementally with high accuracy without any prior assumption about the data sets. To keep an up-to-date fuzzy rule base, a novel `gradual´-forgetting-based rule pruning approach is proposed to unlearn outdated data by deleting obsolete rules. Experiments conducted on real-life traffic data confirm the validity of the design and the accuracy performance of the GSETSK system.