ECOC (Error-Correcting Output Codes) are used to address multi-label classification problems. tECOC (time-windowed ECOC) is an adaptation of ECOC aimed at multivariate time-series classification. It is useful for building fast and (fairly) accurate models which leverage the simplicity and efficiency of two-label classifiers such as linear-discriminant analysis classifiers or support-vector machines.

The repository provides two examples to start the user off (small subsets from EEG and multi-sensor WSN datasets). Any contributions or enhancements are more than welcome. I would like to migrate the code from closed-source MATLAB to something open source, preferably Julia. Any contributions towards this are highly encouraged. The code was originally developed for an EEG study - if you use tECOC in your work, please cite this paper.

Reference  Chauhan, T., Jakovljev, I., Thompson, L. N., Wuerger, S. M., & Martinovic, J. (2023). Decoding of EEG signals reveals non-uniformities in the neural geometry of colour. NeuroImage, 268, 119884. [doi]