The previous post was fortunately wrong. The data transform wasn’t correct. That’s what you get for using a powerful data transformation engine without validating the transformations.. Anyway, the nBack error biases are looking much better..

(more…)