I am a postdoctoral researcher in Mathematical Statistics at Lund University. I am especially interested in theories and methods in
- time-frequency analysis
- machine learning
- non-stationary stochastic processes
- neuroscience applications
R. Anderson and M. Sandsten, “Time-frequency feature extraction for classification of episodic memory” EURASIP Journal on Advances in Signal Processing, Vol. 19, 2020. DOI: 10.1186/s13634-020-00681-8 (OPEN ACCESS)
R. Anderson, M. Sandsten “Multitaper Spectral Granger Causality with Application to SSVEP”, IEEE Proceedings of the 45th International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2020. DOI: 10.1109/ICASSP40776.2020.9054388
M. Sandsten, R. Anderson, I. Reinhold and J. Brynolfsson, “The Matched Reassigned Cross-Spectrogram for Phase Estimation”, IEEE Proceedings of the 45th ICASSP, 2020. DOI: 10.1109/ICASSP40776.2020.9053684
R. Anderson, P. Jönsson and M. Sandsten, “Stochastic Modeling and Optimal Spectral Estimation of Task-Related HRV”, Applied Sciences, Vol. 9, Iss. 23, 5154, 2019. DOI: 10.3390/app9235154 (OPEN ACCESS)
R. Anderson and M. Sandsten, “Inference for Time-varying Signals using Locally Stationary Processes”, Journal of Computational and Applied Mathematics, Vol. 347, p. 24-35, 2019. DOI: 10.1016/j.cam.2018.07.046 (OPEN ACCESS)
U. Picchini and R. Anderson, “Approximate maximum likelihood estimation using data-cloning ABC”, Computational Statistics & Data Analysis, Vol. 105, p. 166-183, 2017. DOI: 10.1016/j.csda.2016.08.006
Master's thesis proposals
If you are a student looking for a Master Thesis project, you may find interesting one of the following proposals.
Measures for brain functional connectivity
Several brain regions cooperate when the human brain is completing a task. This cooperation is called functional connectivity, i.e., the temporal coincidence of spatially distant neurophysiological events. Estimating functional connectivity from EEG measurements is difficult because the signals are very noisy due to volume conduction. This project aims to investigate different functional connectivity measures (e.g., coherence, phase coupling, phase-locking value) using simulated EEG data and test their robustness against varying levels of signal-to-noise ratios.
Prerequisites: FMSF10/MASC04, FMSN45/MASM17
Classification of brain signals using power spectral density estimators and Riemannian geometry
Machine learning methods based on Riemannian geometry for the classification of EEG signals are attracting increasing attention . The features are derived from the measured EEG signals' spatial covariance estimated with the sample covariance matrix, which has several limitations. An alternative approach consists in using the power spectral density . This master thesis project aims to test the performance of different spectral estimators in Riemannian geometry-based classification of EEG data.
 M. Congedo, A. Barachant, and R. Bhatia. "Riemannian geometry for EEG-based braincomputer interfaces; a primer and a review". Brain-Computer Interfaces, Taylor & Francis, 4 (3), pp.155-174, 2017
 Y. Li and K.M. Wong, "Riemannian Distances for Signal Classification by Power Spectral Density", IEEE Journal on Selected Topics in Signal Processing, 2013