Parallel tempering is efficient for learning restricted Boltzmann machines.

KyungHyun Cho, Tapani Raiko, Alexander Ilin

Browse the full IJCNN paper archive.