Recent Research
Learn more about some of our recent research results, supplement material and MATLAB code.
- Ear-EEG, scalp-EEG, sleep, fatigue and Brain Computer Interface
- Widely Linear Complex Valued Adaptive Filtering (linear, Kalman, ESN, neural)
- Quaternion valued signal processing, neural networks, and machine learning
- Signal Processing for Smart Grid and Renewable Energy (mostly 2D and 3D wind prediction)
- Blind Source Separation and Extraction (BSS and BSE), Complex BSS for Noncircular data, Quaternion BSS, Blind Widely Linear Modelling
- Multiscale modelling and Time-Frequency analysis: Multivariate EMD, Multivariate Synchrosqueezing.
- Multivariate Multiscale Entropy and Dynamical Complexity of Multichannel Data
- Surrogate data, hypothesis testing, signal modality characterization, embedding dimension, delay vector variance (DVV) method
- Recurrent Neural Networks
- Data and Sensor Fusion
- Other
- Technical Reports
Current research projects
Nonlinear Multidimensional Adaptive Modeling, System Identification and Prediction — Especially complex-valued and hyper-complex neural networks for natural phenomena, such as in modeling of environmental signals, complex industrial plants and trajectory tracking.
Wind Modeling and Prediction — Using advanced nonlinear multivariate predictors. Simultaneous modeling of all the components of multidimensional and multivariate signals.
One- and multi-dimensional adaptive denoising — Using adaptive signal processing algorithms and wavelets, with applications to biomedical signals, such as fMRI and medical images.
Blind Source Separation and Extraction — Especially on-line adaptive algorithms for Electroencephalogram, MEG, and fMRI, especially based on some fundamental properties of signals, such as their nonlinear, stochastic or nonstationary nature.
Brain Signal Processing — Signal conditioning and classification in various EEG based mental tasks, such as decision making processes.
Sensor Fusion — Combining information from sensors measuring different physical quantities, especially in modeling and detection of sleep stages, where a combination of EEG, EOG, ECG and respiratory signal is processed. Also sensor fusion in navigation, and robotics.
Fit for Duty — Data fusion based techniques for fatigue detection.
Signal Modality Characterisation — Based on the recently developed DVV method, this new exciting area of research can be used to asses a qualitative performance of adaptive filters as well as for initial characterisation of signal's nature, as for instance, linear, nonlinear, deterministic or stochastic.