Data Driven Fuzzy and Neural Dynamic Systems Modeling

The paper addresses data driven fuzzy modeling and long-short term memory structures, and evaluate their performance in modeling nonlinear dynamic systems. These are representative and powerful modeling paradigms representative of the state of the art of fuzzy and deep neural modeling approaches. Models complexity and approximation capabilities are evaluated, and their usefulness is discussed in front of the predictions they produce. Overall, the results suggest that they are competitive with alternative methods such as convolutional and Gaussian process.