Supervised Multiple Timescale Recurrent Neuron Network Model for Human Action Classification

Multiple time-scales recurrent neural network MTRNN model is a useful tool to record and regenerate a continuous signal for a dynamic task. However, the MTRNN itself cannot classify different motions because there are no output nodes for classification tasks. Therefore, in this paper, we propose a novel supervised model called supervised multiple time-scales recurrent neural network SMTRNN to handle the classification issue. The proposed SMTRNN can label different kinds of signals without setting the initial states. SMTRNN provided both prediction and classification signals simultaneously during testing. In addition, the experiment results show that SMTRNN successfully classifies a continuous signal including multiple kinds of actions as well predicts motions.

[1]  Ronald Poppe,et al.  A survey on vision-based human action recognition , 2010, Image Vis. Comput..

[2]  Qing Chen,et al.  Dynamic Gesture Recognition , 2005, 2005 IEEE Instrumentationand Measurement Technology Conference Proceedings.

[3]  Jun Tani,et al.  Achieving "organic compositionality" through self-organization: Reviews on brain-inspired robotics experiments , 2008, Neural Networks.

[4]  Kenji Doya,et al.  Adaptive neural oscillator using continuous-time back-propagation learning , 1989, Neural Networks.

[5]  Shigeki Sugano,et al.  Imitating others by composition of primitive actions: A neuro-dynamic model , 2012, Robotics Auton. Syst..

[6]  Minho Lee,et al.  Neuro-robotics study on integrative learning of proactive visual attention and motor behaviors , 2012, Cognitive Neurodynamics.

[7]  Jun Tani,et al.  Emergence of Functional Hierarchy in a Multiple Timescale Neural Network Model: A Humanoid Robot Experiment , 2008, PLoS Comput. Biol..

[8]  O. Hikosaka,et al.  Chunking during human visuomotor sequence learning , 2003, Experimental Brain Research.

[9]  Tanja Schultz,et al.  HMM-based human motion recognition with optical flow data , 2009, 2009 9th IEEE-RAS International Conference on Humanoid Robots.

[10]  Luiz Velho,et al.  Kinect and RGBD Images: Challenges and Applications , 2012, 2012 25th SIBGRAPI Conference on Graphics, Patterns and Images Tutorials.

[11]  Peter Stagge,et al.  Recurrent neural networks for time series classification , 2003, Neurocomputing.

[12]  Reza Shadmehr,et al.  Learning of action through adaptive combination of motor primitives , 2000, Nature.

[13]  Yuichi Nakamura,et al.  Approximation of dynamical systems by continuous time recurrent neural networks , 1993, Neural Networks.

[14]  Tetsuya Ogata,et al.  Emergence of hierarchical structure mirroring linguistic composition in a recurrent neural network , 2011, Neural Networks.