Design of the high-order sequence prediction task. (A). Structure of high-order
sequences with shared sub-sequences. (B). High-order sequences with multiple possible
endings. (C). Stream of sequences with noise between sequences. Both learning and
testing occur continuously. After the model learned one set of sequences, we switched
to a new set of sequences with contradictory endings to test the adaptation to changes
in the data stream.
This site uses cookies. By continuing to use our website, you are agreeing to our privacy policy. No content on this site may be used to train artificial intelligence systems without permission in writing from the MIT Press.