Decomposition of Chaotic Signals by Reservoir Computing
Here I present my honours thesis, an integral part of the final year of my BSc (hons) in Physics.
In my research I study the evolution of artificial neural networks throughout history and discuss a few of the most prominent structures and learning methods.
After this their is an introduction to the echo state network desgined by Herbert Jaeger. I then move on to analyisng the inner workings of this network implementation, and then demonstrate the utility of this network in modelling a non-linear dynamical system and decomposing linearly combined chaotic signals.