TY - GEN
T1 - A fast training algorithm of multiple-timescale recurrent neural network for agent motion generation
AU - Yu, Zhibin
AU - Mallipeddi, Rammohan
AU - Lee, Minho
N1 - Publisher Copyright:
© 2015 ACM.
PY - 2015/10/21
Y1 - 2015/10/21
N2 - Motion understanding and regeneration are two basic aspects of human-agent interaction. One important function of agents is to represent human's activities. For better interaction with human, robot agents should not only do something following human's order, but also be able to understand or even play some actions. Multiple Timescale Recurrent Neural Networks (MTRNN) is believed to be an efficient tool for robots action generation. In our previous work, we extended the concept of MTRNN and developed Supervised MTRNN for motion recognition. In this paper, we use Conditional Restricted Boltzmann Machine (CRBM) to initialize Supervised MTRNN and accelerate the training speed of Supervised MTRNN. Experiment results show that our method can greatly increase the training speed without losing much performance.
AB - Motion understanding and regeneration are two basic aspects of human-agent interaction. One important function of agents is to represent human's activities. For better interaction with human, robot agents should not only do something following human's order, but also be able to understand or even play some actions. Multiple Timescale Recurrent Neural Networks (MTRNN) is believed to be an efficient tool for robots action generation. In our previous work, we extended the concept of MTRNN and developed Supervised MTRNN for motion recognition. In this paper, we use Conditional Restricted Boltzmann Machine (CRBM) to initialize Supervised MTRNN and accelerate the training speed of Supervised MTRNN. Experiment results show that our method can greatly increase the training speed without losing much performance.
KW - Action generation
KW - Machine learning
KW - Recurrent Neural Network
KW - Restricted Boltzmann Machine
UR - https://www.scopus.com/pages/publications/84962828799
U2 - 10.1145/2814940.2814986
DO - 10.1145/2814940.2814986
M3 - Conference contribution
AN - SCOPUS:84962828799
T3 - HAI 2015 - Proceedings of the 3rd International Conference on Human-Agent Interaction
SP - 243
EP - 246
BT - HAI 2015 - Proceedings of the 3rd International Conference on Human-Agent Interaction
PB - Association for Computing Machinery, Inc
T2 - 3rd International Conference on Human-Agent Interaction, HAI 2015
Y2 - 21 October 2015 through 24 October 2015
ER -