Interspeech'2005 - Eurospeech

Lisbon, Portugal
September 4-8, 2005

Multi-Task Learning Strategies for a Recurrent Neural Net in a Hybrid Tied-Posteriors Acoustic Model

Jan Stadermann, Wolfram Koska, Gerhard Rigoll

Technische Universität München, Germany

An important goal of an automatic classifier is to learn the best possible generalization from given training material. One possible improvement over a standard learning algorithm is to train several related tasks in parallel. We apply the multi-task learning scheme to a recurrent neural network estimating phoneme posterior probabilities and HMM state posterior probabilities, respectively. A comparison of networks with different additional tasks within a hybrid NN/HMM acoustic model is presented. The evaluation has been performed using the WSJ0 speaker independent test set with a closed vocabulary of 5000 words and shows a significant improvement compared to a standard hybrid acoustic model if gender classification is used as additional task.

Full Paper

Bibliographic reference.  Stadermann, Jan / Koska, Wolfram / Rigoll, Gerhard (2005): "Multi-task learning strategies for a recurrent neural net in a hybrid tied-posteriors acoustic model", In INTERSPEECH-2005, 2993-2996.