JAR14

Some minor bug fixes and updates:

• Ability to plot results symmetrical across the diagonal
• Fixes related to computing labels when test set has no labels
• Include SVM (not recommended, very slow)

JAR13

An small but annoying bug was introduced on December 8, which is fixed in this release. When plotting multiple comparison corrected temporal generalization matrices, the axes of the significant time values were accidentally swapped, such that training and testing time for the contour lines were swapped. This has now been fixed. When you do not look at temporal generalization, this bug is not relevant for you.

JAR12

Just fixing a small but relevant bug: when doing time-frequency decoding, no proper electrode selection was taking place, such that these analyses were based on all electrodes rather than a selection. Now fixed. Also renaming the release to JAR12 to be in line with versioning.

JAR9

• Silly, but yet another release after yesterday's, just to improve the plotting of significant time points in temporal generalization matrices. Significant points that show negative decoding are now encircled by dark blue, while ones that show positive decoding are encircled by red.

JAR8

• Fixed/updated trial-masked robust detrending procedure
• IMPORTANT NOTE: A change was made to the plotting procedure of temporal generalization matrices and frequency plots. All uncorrected significant test points are now plotted as saturated points, regardless of the multiple comparison correction method. Cluster-based permutation corrected tests or FDR corrected tests are encircled by a dark red contour line. If no such line appears, none of your tests survives multiple comparison correction.

JAR7

fixed a small bug related to Matlab 2019 that does not allow the usage of variable names that are also the names of functions if they are not explicitly initialized.

JAR6

Some bugs were fixed, in particular one in the adam_detrend_and_epoch. m function. Also, an earlier release contained a function that could not be used on older MATLAB versions. Now fixed.

JAR5

This release is intended to accompany the book chapter "Multivariate methods to track the spatiotemporal profile of attentional feature-based selection using EEG" to appear in "Spatial learning and attention guidance." edited by Stefan Pollmann

JAR4

• I'm sorry to say that a silly error crept into JAR3 with respect to the timeline. This should now be fixed.
• Some more work was done on adam_detrend_and_epoch function, but this work is ongoing

JAR3

• Improved help and support for forward encoding models (FEMs)
• Some work was done on correlating the confidence scores of the classifier, but poorly validated
• Added a new function to properly remove drifts from EEG using polynomials fits rather than highpass filtering: adam_detrend_and_epoch
• Many small updates and bug fixes
• As always, the software is in permanent beta status