Version history

• Release 09-12-2007, version 1.3: minor changes.
  • SOBI is again the default algorithm for BSS-based EOG correction.
  • Documentation has been updated and correction examples have been included.
• Release 04-12-2007, version 1.3: minor bugs and few major bugs corrected.
  • A few minor and major bugs were still left in the BSS interface functions and have now been corrected.
  • The interface functions use now pinv instead of inv to avoid numerically unstable results when the data covariance matrix is close to singular. This numerical unstability might be the reason for the small differences that were observed when running the artifact correction methods under different MATLAB versions and under different operating systems.
  • The default window length in the automatic EMG correction method is now twice as long as it was before.

• Release 03-12-2007, version 1.3: major bugs corrected.
  • Several major bugs have been corrected in the functions that act as interface to the BSS algorithms.
  • Parameters passed to the BSS algorithms were ignored in previous releases due to a bug in function autobss.m. This has been corrected.

• Release 29-11-2007, version 1.3: minor bugs corrected.
  • Minor bugs corrected.

• Release 28-11-2007, version 1.3: major bugs corrected.
  • A major bug in function emg_psd.m has been corrected. This bug was causing the emg_psd.m criterion to incorrectly detect the EMG-related components.
  • Minor update of function pop_autobssemg.m to adapt to the new version of function emg_psd.m
  • The criterion emg_psd.m now accepts a new parameter that allows the user to select the spectral estimator to be used to compute the EEG average power and the EMG average power of each estimated component.
  • A compatibility issue related to function emg_psd.m has been solved. Previously to this change, the criterion emg_psd was producing different results when using MATLAB's Signal Processing Toolbox 6.6 and previous when using older versions of the Signal Processing Toolbox. Now it produces ''almost'' the same results.
  • Several other minor bugs corrected.

• Release 07-11-2007, version 1.3: minor update.
  • pop_autobssemg now automatically passes the sampling rate to emg_psd.m
  • First draft of the toolbox documentation.

• Release 31-10-2007, version 1.3: minor update.
  • Minor bugs corrected.
  • EFICA v1.9 was updated with EFICA v2.0 (implemented by Z. Koldovský).

• Release 29-10-2007, version 1.3: iWASOBI, EFICA, COMBI, FCOMBI added.
  • Few minor bugs corrected.
  • Slightly improved documentation.
  • A sample EEG dataset has been included in the release.
  • Added several new algorithms for BSS: iWASOBI, EFICA, COMBI, FCOMBI.

• Release 01-07-2006, version 1.2:, Regression-based and PCA-based EOG correction added.
  • Automatic EOG correction using Least Mean Squares (LMS) adaptive filtering. This is implemented in lms_regression.m.
  • Automatic EOG correction using the conventional Recursive Least Squares (RLS) adaptive algorithm. This is implemented in function crls_regression.m.
  • Automatic EOG correction using an stable version of the RLS algorithm. This is implemented in scrls_regression.m.
  • Automatic EOG correction using two H-infinity regression methods. These are implemented in functions hinftv_regression.m and hinfew_regression.m.
  • Automatic EOG correction using Principal Component Analysis (PCA).
  • Some bugs corrected.

• Release 10-04-2006, version 1.1: EMG correction included.
  • Automatic correction of EMG artifacts using canonical correlation analysis. This is implemented in functions bsscca.m and emg_psd.m.
  • Function BSS has been modified to separate the actual BSS algorithm from the components selection criteria.
  • Interfaces to BSSCCA, FastICA, JADE and RUNICA (implementation of Infomax included in the EEGLAB toolbox) have been included.
  • Ill-conditioning of the data covariance matrix is now handled properly.
  • Some minor bugs corrected.

• Release 27-02-2006, version 1.0: original release.
  • Automatic correction of EOG artifacts using Blind Source Separation (BSS).
  • Only an interface to SOBI is provided.