Leyuan Liu

Dr. Leyuan Liu

Faculty of Information Technology, Beijing University of Technology, Beijing 100124, China
orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 July 2020 in Sensors
Verified
Journal Article
Open access
PDF file
Leyuan LiuLeyuan Liu
Leyuan Liu
YHYibin Hou
Yibin Hou
JHJian He
Jian He
JLJonathan Lungu
Jonathan Lungu
RDRuihai Dong
Ruihai Dong
A fall detection module is an important component of community-based care for the elderly to reduce their health risk. It requires the accuracy of detections as well as maintains energy saving. In order to meet the above requirements, a sensing module-integrated energy-efficient sensor was developed which can sense and cache the data of human activity in sleep mode, and an interrupt-driven algorithm is proposed to transmit the data to a server integrated with ZigBee. Secondly, a deep neural network for fall detection (FD-DNN) running on the server is carefully designed to detect falls accurately. FD-DNN, which combines the convolutional neural networks (CNN) with long short-term memory (LSTM) algorithms, was tested on both with online and offline datasets. The experimental result shows that it takes advantage of CNN and LSTM, and achieved 99.17% fall detection accuracy, while its specificity and sensitivity are 99.94% and 94.09%, respectively. Meanwhile, it has the characteristics of low power consumption.

6 December 2021 in Entropy
Verified
Journal Article
Open access
PDF file
Leyuan LiuLeyuan Liu
Leyuan Liu
JHJian He
Jian He
KRKeyan Ren
Keyan Ren
JLJonathan Lungu
Jonathan Lungu
YHYibin Hou
Yibin Hou
RDRuihai Dong
Ruihai Dong
Wearable sensor-based HAR (human activity recognition) is a popular human activity perception method. However, due to the lack of a unified human activity model, the number and positions of sensors in the existing wearable HAR systems are not the same, which affects the promotion and application. In this paper, an information gain-based human activity model is established, and an attention-based recurrent neural network (namely Attention-RNN) for human activity recognition is designed. Besides, the attention-RNN, which combines bidirectional long short-term memory (BiLSTM) with attention mechanism, was tested on the UCI opportunity challenge dataset. Experiments prove that the proposed human activity model provides guidance for the deployment location of sensors and provides a basis for the selection of the number of sensors, which can reduce the number of sensors used to achieve the same classification effect. In addition, experiments show that the proposed Attention-RNN achieves F1 scores of 0.898 and 0.911 in the ML (Modes of Locomotion) task and GR (Gesture Recognition) task, respectively.

1 March 2021
Verified
Conference Paper
Closed access
ZAZhang Aiwen
Zhang Aiwen
Leyuan LiuLeyuan Liu
Leyuan Liu
With the development of mobile Internet era, mobile terminal equipment has become an indispensable tool in people's daily life. In order to improve the user experience and ease the endurance of mobile terminals, this paper studies the task offloading strategy for optimizing the energy consumption of mobile terminals. At present, most of the research only focuses on a single cloud offload and edge offload, and the research on reducing the energy consumption of mobile terminals is insufficient. In order to further reduce the energy consumption of mobile terminal equipment and improve the utilization of computing resources, this paper proposes a cloud-edge-terminal multi-computing resource cooperation task offloading model, and then designs a fitness calculation method for this model, and proposes a multi-computing resources cooperation energy optimization algorithm based on particle swarm optimization algorithm. Experiments show that, compared with the existing four offloading strategies, the proposed algorithm can fully reduce the energy consumption of mobile terminal devices while meeting the requirements of user delay, and has advantages in large-scale task offloading.

1 July 2020
Verified
Conference Paper
Closed access
Leyuan LiuLeyuan Liu
Leyuan Liu
JHJian He
Jian He
YHYibin Hou
Yibin Hou
CZCheng Zhang
Cheng Zhang
The bus passenger data are very important for urban bus dispatching management. When passengers get on or off the bus, they often hide from each other. It is a great challenge for automatically accounting passengers through camera. The traditionally video-based target detection algorithm or target tracking algorithm is difficult to accurately count the number of passenger on and off. In this paper, the YOLOv2 algorithm is combined with the MIL tracker so as to real-time account the number of passengers in the bus surveillance video. Experiment shows that the accuracy rate of bus passenger statistics proposed in this paper reaches over 99%, and it proves that our method has good real-time and high accuracy.

1 October 2019
Verified
Conference Paper
Closed access
TNTong Nie
Tong Nie
Leyuan LiuLeyuan Liu
Leyuan Liu
DSDa Sun
Da Sun
In the past few decades, "explicit" and "implicit" feedback information in recommendation systems have been researched as two separate areas. But, in the real world, user behaviors contain a series of explicit and implicit information, ranging from clicks, collections to purchases and ratings. It is still an unsolved problem of how to effectively use abundant implicit signals to help predict more accurate explicit signals and thus generate more appropriate recommendations. Therefore, we propose a multiple monotonic behavior chains model (MC-Rec), which can simulate the full range of interactions. The Adaboost algorithm is used to set the weights of each behavior chain, and the monotony of the behavior chain is lucubrated. The experimental results are evaluated in three real data sets, and demonstrate that MC-Rec model can effectively improve the prediction accuracy compared with the state-of-the-art baselines.