TY - GEN
T1 - Fully Convolutional Network Bootstrapped by Word Encoding and Embedding for Activity Recognition in Smart Homes
AU - Bouchabou, Damien
AU - Nguyen, Sao Mai
AU - Lohr, Christophe
AU - LeDuc, Benoit
AU - Kanellos, Ioannis
N1 - Publisher Copyright:
© 2021, Springer Nature Singapore Pte Ltd.
PY - 2021/1/1
Y1 - 2021/1/1
N2 - Activity recognition in smart homes is essential when we wish to propose automatic services for the inhabitants. However, it is a challenging problem in terms of environments’ variability, sensory-motor systems, user habits, but also sparsity of signals and redundancy of models. Therefore, end-to-end systems fail at automatically extracting key features, and need to access context and domain knowledge. We propose to tackle feature extraction for activity recognition in smart homes by merging methods of Natural Language Processing (NLP) and Time Series Classification (TSC) domains. We evaluate the performance of our method with two datasets issued from the Center for Advanced Studies in Adaptive Systems (CASAS). We analyze the contributions of the use of embedding based on term frequency encoding, to improve automatic feature extraction. Moreover we compare the classification performance of Fully Convolutional Network (FCN) from TSC, applied for the first time for activity recognition in smart homes, to Long Short Term Memory (LSTM). The method we propose, shows good performance in offline activity classification. Our analysis also shows that FCNs outperforms LSTMs, and that domain knowledge gained by event encoding and embedding improves significantly the performance of classifiers.
AB - Activity recognition in smart homes is essential when we wish to propose automatic services for the inhabitants. However, it is a challenging problem in terms of environments’ variability, sensory-motor systems, user habits, but also sparsity of signals and redundancy of models. Therefore, end-to-end systems fail at automatically extracting key features, and need to access context and domain knowledge. We propose to tackle feature extraction for activity recognition in smart homes by merging methods of Natural Language Processing (NLP) and Time Series Classification (TSC) domains. We evaluate the performance of our method with two datasets issued from the Center for Advanced Studies in Adaptive Systems (CASAS). We analyze the contributions of the use of embedding based on term frequency encoding, to improve automatic feature extraction. Moreover we compare the classification performance of Fully Convolutional Network (FCN) from TSC, applied for the first time for activity recognition in smart homes, to Long Short Term Memory (LSTM). The method we propose, shows good performance in offline activity classification. Our analysis also shows that FCNs outperforms LSTMs, and that domain knowledge gained by event encoding and embedding improves significantly the performance of classifiers.
KW - Automatic features
KW - Embedding
KW - Fully Convolutional Network
KW - Human activity recognition
KW - Smart homes
KW - Word encoding
U2 - 10.1007/978-981-16-0575-8_9
DO - 10.1007/978-981-16-0575-8_9
M3 - Conference contribution
AN - SCOPUS:85102756809
SN - 9789811605741
T3 - Communications in Computer and Information Science
SP - 111
EP - 125
BT - Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings
A2 - Li, Xiaoli
A2 - Wu, Min
A2 - Chen, Zhenghua
A2 - Zhang, Le
PB - Springer Science and Business Media Deutschland GmbH
T2 - 2nd International Workshop on Deep Learning for Human Activity Recognition, DL-HAR 2020, held in conjunction with IJCAI-PRICAI 2020
Y2 - 8 January 2021 through 8 January 2021
ER -