Add character associated to each trial in P300 datasets

[qbarthelemy]

In P300 datasets, each trial is usually associated to a (binary) ERP class : target (target character has been flashed, evoking an ERP in the trial) versus non-target (target character has not been flashed, no ERP in trial).

But, we ignore the character class, ie the label of this target character, among all possible characters of the interface.

In P300 datasets, labels should contain:

• the ERP class for each trial;
• the list of flashed characters for each trial;
• the character class (ie, the target character) for each group of trials.

[sylvchev]

Yes, we have discussed this during office hours. @v-goncharenko is the the case for the Neiry Demon dataset? Your classification methods rely on this information as well?

[v-goncharenko]

Yes, this is the case we've discussed. You proposed to add different paradigm for that case.

Also this character class could be inferred from existing datasets e.g. I did that for Brain Invaders in my private repo

[qbarthelemy]

@v-goncharenko , can you explain how to retrieve character classes from any P300 dataset? and for Brain Invaders, can you commit them into public repo? thx a lot!