A Memristive-Friendly Recurrent Neural Network model for time-series processing

This thesis introduces the Memristive-Friendly Echo State Network (MF-ESN) and Recurrent Neural Network (MF-RNN) as computational neural dynamical models inspired by memristive materials. It analyzes their dynamics and benchmarks their potential for complex sequence-based applications. This work bridges Reservoir Computing and in-materia computing, promising neuromorphic model advancement.