Method #1 (David Pinto) Uses a Hidden Markov Model with 10 states. result for data 1A in Method1.txt Method #2 (Justin Sanchez, Deniz Erdogmus, Tue Lehn-Schioeler, Yadu Rao) Recursive Multi-Layer Perceptron with fully-connected MLP with 5 hidden processing elements. result for data 1A in Method2.txt Method #3 (Kenneth Hild, Tue Lehn-Schioeler) Temporal Principle Components Analysis preprocessing (uses the three largest, length-five eigenvectors as features, for a total of 90 features), followed by information-theoretic feature reduction (reducing the number of features to 10), followed by a non-parametric Bayes classifier. result for data 1A in Method3.txt Method #4 (Yadu Rao, David Pinto) Time-delay neural network predictor using an embedding of 5 (30 x 8 x 6). result for data 1A in Method4.txt Method #5 (Deniz Erdogmus, Yadu Rao, David Pinto, Kenneth Hild, Tue Lehn-Schioeler, Justin Sanchez) Mixture of experts of 5 methods, the four above and one other that were similar to one of the above, using a majority vote. result for data 1A in Method5.txt