Deep Convolutional Neural Networks on Multichannel Time Series for Human Activity Recognition

By Jianbo Yang, Minh Nhut Nguyen, Phyo Phyo San, Xiaoli Li, Shonali Priyadarsini Krishnaswamy. International Joint Conference on Artificial Intelligence, 2015

1,181 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/39a1abcbb87d8ff48ee47d446411a3455451f25b

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Classification of time-series images using deep convolutional neural networks (2017, methodology): “22 A sliding window strategy is adopted to put time-series segments into a collection of short pieces of signals.”
Recent trends in machine learning for human activity recognition—A survey (2018, methodology): “The traditional method (shallow learning) requires feature engineering from the data, which is heuristic driven and heavily dependent on human knowledge of the domain (Yang, Nguyen, San, Li, & Krishnaswamy, 2015).”
Deep Learning in Human Activity Recognition with Wearable Sensors: A Review on Advances (2021, methodology): “Self-reflection and self-regulation of one’s own behavior and habit has been an important factor in designing interventions that prevent chronic diseases such as obesity, diabetes, and cardiovascular diseases.”
Ensembles of Deep LSTM Learners for Activity Recognition using Wearables (2017, background): “A multitude of applications are based on CNNs, including but not limited to [Zeng et al. 2014; Ronaoo and Cho 2015; Yang et al. 2015; Rad et al. 2015].”
IoT Wearable Sensor and Deep Learning: An Integrated Approach for Personalized Human Activity Recognition in a Smart Home Environment (2019, methodology): “In order to choose which architecture could reach the best results in our context, a dataset has been collected with the proposed device and preliminary experiments have been carried out on a LSTM [40] and on a CNN [36] already presented in literature.”
Modeling Long- and Short-Term Temporal Patterns with Deep Neural Networks (2017, methodology): “tic categories. RNN has also been applied to mobile data, for classifying the input sequences with respect to actions or activities [13]. CNN models have also been used in action/activity recognition [13, 20, 31], for the extraction of shift-invariant local patterns from input sequences as the featu”
A Semisupervised Recurrent Convolutional Attention Model for Human Activity Recognition (2020, background): “Most existing works follow a supervised learning approach [3], [4], thus requiring a significant amount of training data to recognize meaningful activities.”
Deep Air Quality Forecasting Using Hybrid Deep Learning Framework (2018, background): “CNN is very popular for image processing and target recognition [10], and it is also successfully applied to time series forecasting tasks [24], due to the one-dimensional structure of single time series and two-dimensional structure of multivariate time series.”
Mobile sensor data anonymization (2018, background): “e users. Feature maps of a Convolutional Auto-Encoder (CAE) have the ability to extract patterns and dependencies among data points of a vector and have shown good performance in time series analysis [36]. CAE is usually trained by minimizing the differences (e.g. mean squared error or cross entropy”
Deep Residual Bidir-LSTM for Human Activity Recognition Using Wearable Sensors (2017, methodology): “ F1 score with the NULL class of each algorithm with the Opportunity data set Algorithm F1 score LDA [7] 69% QDA [7] 53% NCC [7] 51% 1NN [7] 87% 3NN [7] 85% UP [7] 64% NStar [7] 84% SStar [7] 86% DBN [9] 73.0% CNN [9] 85.1% Deep-Res-Bidir-LSTM 90.20% It can be seen that the proposed algorithm outper”
Protecting Sensory Data against Sensitive Inferences (2018, methodology): “At the cloud side the Estimator (MTCNN) is trained on a public dataset, the MobiAct dataset in our case.”
Human Activity Recognition using Inertial, Physiological and Environmental Sensors: A Comprehensive Survey (2020, background): “mental sensors (temperature [7], [60], [76], [79], [87]–[89], humidity [79], [87], light [60], [90], Passive Infrared Sensor (PIR) [88]), radio signals (WiFi and Bluetooth [56], [62], [87]), medical equipment (ECG [56], [77], EMG [72]) or other type of build in sensors (GPS [7], [43],”
A Review of Deep Learning Models for Time Series Prediction (2019, methodology): “To the best of our knowledge, we found that more works utilize the CNN architectures for classiﬁcation rather than prediction of the next value with multivariate time-series, such as predicting heart failure from heartbeat data [73] or activity from biometric time series [75].”