During this seminar talk, a brief overview of evolving cascade neuro-fuzzy architectures and their learning algorithms for hybrid cascade neural networks with pool optimization in every cascade will be given. These systems are mainly different from existing cascade systems in its capability to operate in an online mode, which allows it to work with non-stationary and stochastic non-linear chaotic signals with the required accuracy. Some different types of nodes for these systems as well as some pool (ensemble) optimization procedures will be considered. Compared to conventional analogs, these evolving systems provide computational simplicity and possess both tracking and filtering capabilities. Some attention will be also paid to a practical side of application of the introduced systems.