Article
Engineering, Multidisciplinary
Guanghui Wen, Jian Qin, Xingquan Fu, Wenwu Yu
Summary: In this article, a distributed long short-term memory (DLSTM) neural networks is proposed and deployed on IoT devices to handle large-scale spatiotemporal correlation regression tasks. The DLSTM neural networks adopt a collaborative computing architecture with terminals, edges, and cloud to achieve lightweight deep learning and improve learning efficiency. The introduction of distributed memory cells and attention mechanism enhances the generalization ability of LSTM neural networks, while deep fully connected networks deployed on the cloud extract spatiotemporal correlations for improved transferability.
IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING
(2022)
Article
Chemistry, Multidisciplinary
Nhat-Minh Dang-Quang, Myungsik Yoo
Summary: This paper proposes a system architecture based on Kubernetes with a proactive custom autoscaler using a deep neural network model to handle workload dynamically. Experimental results show that the Bi-LSTM model outperforms other models in terms of accuracy and speed of prediction. Furthermore, the proposed autoscaler performs better than Kubernetes' default Horizontal Pod Autoscaler in terms of accuracy and speed of resource provisioning and de-provisioning.
APPLIED SCIENCES-BASEL
(2021)
Article
Automation & Control Systems
Amir Bidokhti, Shahrokh Ghaemmaghami
Summary: This paper introduces a graph-based neural memory module that can be trained using differentiable mechanisms to solve tasks with long-term dependencies. Inspired by the human memory system, this module performs better than traditional methods in terms of convergence speed and final error.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Computer Science, Artificial Intelligence
Shiming Xiang, Bo Tang
Summary: The article introduces a variation of DNC architecture called CSLM-DNC, which includes convertible short-term and long-term memory to improve memory efficiency. Inspired by the human brain, this new scheme improves learning performance through different memory locations importance and memory transformation.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2021)
Article
Environmental Sciences
Hua Su, Tianyi Zhang, Mengjing Lin, Wenfang Lu, Xiao-Hai Yan
Summary: This study proposed a new method using Bi-LSTM neural networks to predict global ocean subsurface temperature and salinity anomalies, which combined surface remote sensing observations and longitude and latitude information, showing improved robustness and accuracy. The model was validated for different months in 2010 and 2015, demonstrating superior performance compared to traditional random forest methods in predicting subsurface thermohaline structure from remote sensing data.
REMOTE SENSING OF ENVIRONMENT
(2021)
Article
Green & Sustainable Science & Technology
Elissaios Sarmas, Evangelos Spiliotis, Efstathios Stamatopoulos, Vangelis Marinakis, Haris Doukas
Summary: This paper proposes a meta-learning method to improve short-term deterministic forecasts of PV systems by blending the base forecasts of multiple DL models. Results indicate that different base models perform best at different PV plants, and meta-learning can improve accuracy by up to 5% over the most accurate base model per plant.
Article
Computer Science, Artificial Intelligence
Keyhan Gavahi, Peyman Abbaszadeh, Hamid Moradkhani
Summary: The DeepYield model, which combines ConvLSTM layers with 3DCNN, is proposed for crop yield forecasting. The model is trained using historical data and remote sensing imagery, with comparisons showing significantly better forecasting performance compared to competing approaches in the soybean growing counties in the United States.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Engineering, Electrical & Electronic
Hongwei Guo, Hongyang Bai, Weiwei Qin
Summary: This article proposes a lightweight deep-learning-based framework for cloud detection in remote sensing imagery. The framework combines multiple feature fusion strategy and fully convolutional neural network technology, which can accurately detect clouds in remote sensing images and outperforms other methods in effectiveness and accuracy.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Green & Sustainable Science & Technology
Syed Altan Haider, Muhammad Sajid, Hassan Sajid, Emad Uddin, Yasar Ayaz
Summary: This study utilizes statistical and Deep Learning techniques to forecast solar Global Horizontal Irradiance in Islamabad, Pakistan, aiming to promote renewable energy development for tackling global climate change. The research finds that ANN, CNN, and LSTM perform best for short-term forecasts, while SARIMAX and Prophet are efficient for long-term forecasts.
Article
Computer Science, Artificial Intelligence
Zhuang Ye, Jianbo Yu
Summary: Machine health assessment is crucial for prognostics and health management, and the proposed LSTMCAE demonstrates effectiveness in feature learning and generating health index using multivariate Gaussian distribution. Experimental results show the superiority of LSTMCAE in machine health assessment compared to other unsupervised learning methods.
APPLIED SOFT COMPUTING
(2021)
Article
Computer Science, Interdisciplinary Applications
Haonan He, Liangyu Chen, Shanyong Wang
Summary: Accurate demand forecasting is crucial for airlines to cope with competition and increase revenue. This paper presents adaptive frameworks based on LSTM network for predicting flight booking demand, achieving superior performance and handling irregular data patterns effectively.
COMPUTERS & INDUSTRIAL ENGINEERING
(2023)
Article
Automation & Control Systems
Ning Jin, Yongkang Zeng, Ke Yan, Zhiwei Ji
Summary: Artificial intelligence-based air quality index (AQI) forecasting is a hot research topic, and the proposed multiple nested long short term memory network (MTMC-NLSTM) model performs superior in accurate AQI forecasting.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
Article
Energy & Fuels
Neethu Elizabeth Michael, Shazia Hasan, Ahmed Al-Durra, Manohar Mishra
Summary: Accurate forecasting is crucial for integrating solar renewables and minimizing intermittent effects. This study proposes an optimized deep learning model for predicting solar time series data, which demonstrates high accuracy and reliability.
Article
Computer Science, Artificial Intelligence
Ronit Jaiswal, Girish K. Jha, Rajeev Ranjan Kumar, Kapil Choudhary
Summary: The study developed a deep long short-term memory (DLSTM) based model for accurate forecasting of nonstationary and nonlinear agricultural prices series. The DLSTM model, advantageous in capturing nonlinear and volatile patterns, demonstrated superiority in price forecasting ability.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Jining Yan, Yuanxing Liu, Lizhe Wang, Zhipeng Wang, Xiaohui Huang, Hong Liu
Summary: This study proposes an efficient data organization method for improving retrieval and access efficiencies of large-scale and long time-series remote-sensing data in a cloud-computing environment. It constructs an asymmetrical index model and prepartitioning mechanism to address low retrieval efficiency, and divides remote-sensing images into tiles with consistent hash operations to enhance access efficiency.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)