This repository contains all resources for the study project "Controlling non-autonomous dynamical systems with conceptors - Exploring high-dimensional state spaces". We focus on applying conceptors as first described by Herbert Jäger to birdsong data, generating a hierarchical model that is capable of recognizing and classifying different sequences of syllables.
All project files are written in python 3 and some of the demo files use IPython Notebooks. The dependencies needed for this project can be installed via:
pip3 install numpy python_speech_features
The source files for running the classification and implementation of Conceptors can be found in the src directory, while the files to download birdsongs and syllables reside in the data folder.