Menü
The Robust Data Science Group is committed to reproducible research. An important contribution to making scientific analyses reproducible is sharing well documented codes.
A toolbox for robust signal processing that can be freely used for non-commercial use only. Please make appropriate references to our book: Zoubir, A. M., Koivunen, V., Ollila, E., and Muma, M. Robust Statistics for Signal Processing Cambridge University Press, 2018.
A set of toolboxes on robust cluster analysis using Bayesian cluster enumeration criteria, heavy-tailed mixture models and M-estimation. Please make reference to our works, e.g., F. K. Teklehaymanot, M. Muma, and A. M. Zoubir, „Bayesian Cluster Enumeration Criterion for Unsupervised learning“, IEEE Trans. Signal Proc. 2018, and C. A. Schroth, and M. Muma, „Robust M-Estimation based Bayesian Cluster Enumeration for Real Elliptically Symmetric Distributions“ IEEE Trans. Signal Proc. 2021
The T-Rex Selector: Fast High-Dimensional Variable Selection with False Discovery Rate Control (R)
An R-Package implementing the Terminating-Random Experiments (T-Rex) selector, a fast variable selection method for high-dimensional data. The T-Rex selector controls a user-defined target false discovery rate (FDR) while maximizing the number of selected variables. Please make reference to our work: J. Machkour, M. Muma, and D. P. Palomar, “The Terminating-Random Experiments Selector: Fast High-Dimensional Variable Selection with False Discovery Rate Control”, Preprint at arXiv:2110.06048.
cran
The T-Lars Algorithm: A High-Dimensional Forward Variable Selection Method (R)
An R-Package implementing the Terminating-Lars algorithm. The T-Lars algorithm terminates the Lars algorithm using dummy variables. It is a building block for the T-Rex, and it allows for very fast and high-dimensional variable selection. Please make reference to our work: J. Machkour, M. Muma, and D. P. Palomar, “The Terminating-Random Experiments Selector: Fast High-Dimensional Variable Selection with False Discovery Rate Control”, Preprint at arXiv:2110.06048.
An algorithm for motion artifact removal in ECG signals. Please make reference to our work: F. Strasser, M. Muma, and A. M. Zoubir, “Motion artifact removal in ECG signals using multi-resolution thresholding.” In Proc. European Signal Processing Conference (EUSIPCO) 2012.
A community detection method that uses spectral partitioning based on estimating a robust and sparse graph model. Please make reference to our work: A. Tastan, M. Muma, A. M. Zoubir, “Sparsity-aware Robust Community Detection (SPARCODE)”, Signal Processing 2021.
A clustering algorithm that exploits gravitational force between mass units to determine regions of highest density of feature vectors. Application in camera networks. Please make reference to our work: P. Binder, M. Muma, and A. M. Zoubir, “Gravitational Clustering: A Simple, Robust and Adaptive Approach for Distributed Networks”. Signal Processing, 2018.
An algorithm for multi-speaker voice activity detection and source enumeration in wireless acoustic sensor networks (WASN). Please make reference to our work: T. Hasija, M. Gölz, M. Muma, P. J. Schreier and A. M. Zoubir, „Source Enumeration and Robust Voice ActivityDetection in Wireless Acoustic Sensor Networks,“ Proc. Asilomar 2019.