RIKEN ABSP Lab
The Cichocki Laboratory for
Brain Signal Processing (ABSP) investigates and develops tools and
software for analysis, extraction, enhancement, de-noising, detection,
localization, recognition, and classification of brain signals and
patterns, especially measured by high density array EEG/MEG and fMRI.
The main objective of the laboratory is develop novel multi-way (tensor)
and machine learning (ML) technologies for massive brain (and generally multi-modal biomedical) data analysis
and for computational (neuro)science, that is modeling and simulations of complex mechanisms and phenomena.
We are developing novel algorithms and software for multiway component analysis,
including multilinear Independent Component Analysis (ICA), non-negative matrix/tensor factorization (NMF/NTF),
Smooth Component Analysis and Sparse Component Analysis (SCA). We investigate novel models
and algorithms for tensor decomposition and tensor networks to simulate complex
systems and process massive large-scale multidimensional data sets.
The potential applications of tensor technology (tensor decompositions and tensor networks)
and multiway component analysis are:
Another topics of our research are:
- Feature extraction.
- Dimensionality reduction
- Classification and hierarchical clustering.
- Anomaly detection.
- Multiomadal data fusion.
- Very large scale optimization problems.
- Brain Computer Interface (BCI).
- Human Robot Interactions (HRI).
- EEG hyper-scanning and their application for rehabilitation, learning and therapies.