Authors:
Mubarak G. Abdu-Aguye
1
and
Walid Gomaa
2
Affiliations:
1
Computer Science and Engineering Department, Egypt-Japan University of Science and Technology and Egypt
;
2
Computer Science and Engineering Department, Egypt-Japan University of Science and Technology, Egypt, Faculty of Engineering, Alexandria University and Egypt
Keyword(s):
Activity Recognition, Transfer Learning, IMU, Convolutional Neural Networks, Deep Learning.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Engineering Applications
;
Image and Video Analysis
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Optimization Problems in Signal Processing
;
Robotics and Automation
;
Sensors Fusion
;
Signal Processing, Sensors, Systems Modeling and Control
;
Time-Frequency Analysis
Abstract:
The advent of Deep Learning has, together with massive gains in predictive accuracy, made it possible to reuse knowledge learnt from solving one problem in solving related problems. This is described as Transfer Learning, and has seen wide adoption especially in computer vision problems, where Convolutional Neural Networks have shown great flexibility and performance. On the other hand, transfer learning for sequences or timeseries data is typically made possible through the use of recurrent neural networks, which are difficult to train and prone to overfitting. In this work we present VersaTL, a novel approach to transfer learning for fixed and variable-length activity recognition timeseries data. We train a Convolutional Neural Network and use its convolutional filters as a feature extractor, then subsequently train a feedforward neural network as a classifier over the extracted features for other datasets. Our experiments on five different activity recognition datasets show the pr
omise of this method, yielding results typically within 5% of trained-from-scratch networks while obtaining between a 24-52x reduction in the training time.
(More)