Learning Compact Features for Human Activity Recognition Via Probabilistic First-Take-All
- 8 October 2018
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 42 (1) , 126-139
Abstract
With the popularity of mobile technology, wearable devices, such as smart wristbands and smartphones, open a unprecedented opportunity to solve the challenging human activity recognition (HAR) problem by learning expressive representations from the multi-dimensional sensor signals that record huge amounts of daily activities. This inspires us to develop a new algorithm applicable to both camera-based and wearable sensor-based HAR systems. Although competitive classification accuracy has been reported, existing methods often face the challenge of distinguishing visually similar activities composed of activity patterns in different temporal orders. In this paper, we propose a novel probabilistic algorithm to compactly encode temporal orders of activity patterns for HAR. Specifically, the algorithm learns an optimal set of latent patterns such that their temporal structures really matter in recognizing different human activities. Then a novel probabilistic First-Take-All (pFTA) approach is introduced to generate compact features from the orders of these latent patterns to encode an entire sequence, and the temporal structural similarity between different sequences can be efficiently computed by the Hamming distance between compact features. Experiments on three public HAR datasets show the proposed pFTA approach can achieve competitive performance in terms of accuracy as well as efficiency.Keywords
Funding Information
- National Aeronautics and Space Administration (NNX15AV40A)
- National Science Foundation (1560302)
This publication has 26 references indexed in Scilit:
- Transition-Aware Human Activity Recognition Using SmartphonesNeurocomputing, 2016
- Real-time human pose recognition in parts from single depth imagesCommunications of the ACM, 2013
- Accuracy and Resolution of Kinect Depth Data for Indoor Mapping ApplicationsSensors, 2012
- LIBSVMACM Transactions on Intelligent Systems and Technology, 2011
- Sequential Deep Learning for Human Action Recognition2011
- Comparative study on classifying human activities with miniature inertial and magnetic sensorsPattern Recognition, 2010
- Using acceleration measurements for activity recognition: An effective learning algorithm for constructing neural classifiersPattern Recognition Letters, 2008
- The random subspace method for constructing decision forestsIEEE Transactions on Pattern Analysis and Machine Intelligence, 1998
- Long Short-Term MemoryNeural Computation, 1997
- A desicion-theoretic generalization of on-line learning and an application to boosting1995