
Alexandria Engineering Journal, Год журнала: 2025, Номер 116, С. 586 - 600
Опубликована: Янв. 8, 2025
Alexandria Engineering Journal, Год журнала: 2025, Номер 116, С. 586 - 600
Опубликована: Янв. 8, 2025
Frontiers in Energy Research, Год журнала: 2023, Номер 11
Опубликована: Май 16, 2023
Introduction: Smart grid financial market forecasting is an important topic in deep learning. The traditional LSTM network widely used time series because of its ability to model and forecast data. However, long-term forecasting, the lack historical data may lead a decline performance. This difficult problem for networks overcome. Methods: In this paper, we propose new deep-learning address problem. WOA-CNN-BiLSTM combines bidirectional long short-term memory BiLSTM convolution Advantages Neural Network CNN. We replace with network, BiLSTM, exploit capturing dependencies. It can capture dependencies modelling. At same time, use convolutional neural (CNN) extract features better represent patterns regularity method combining CNN learn characteristics more comprehensively, thus improving accuracy prediction. Then,to further improve performance CNN-BiLSTM model, optimize using whale algorithm WOA. optimization algorithm, which has good global search convergence speed, complete short time. Results: Optimizing through WOA reduce calculation training prediction smart market, market. Experimental results show that our proposed than other models effectively deal missing sequence forecasting. Discussion: provides necessary help development markets risk management services, promote growth industry. Our research are great significance learning, provide effective idea solving grid.
Язык: Английский
Процитировано
9Multimedia Tools and Applications, Год журнала: 2023, Номер 83(11), С. 31379 - 31394
Опубликована: Сен. 16, 2023
Язык: Английский
Процитировано
9Frontiers in Neuroscience, Год журнала: 2023, Номер 17
Опубликована: Сен. 20, 2023
In the medical field, electronic records contain a large amount of textual information, and unstructured nature this information makes data extraction analysis challenging. Therefore, automatic entity from has become significant issue in healthcare domain.To address problem, paper proposes deep learning-based model called Entity-BERT. The aims to leverage powerful feature capabilities learning pre-training language representation BERT(Bidirectional Encoder Representations Transformers), enabling it automatically learn recognize various types records, including terminologies, disease names, drug more, providing more effective support for research clinical practices. Entity-BERT utilizes multi-layer neural network cross-attention mechanism process fuse at different levels types, resembling hierarchical distributed processing human brain. Additionally, employs pre-trained sequence models data, sharing similarities with semantic understanding Furthermore, can capture contextual long-term dependencies, combining handle complex diverse expressions method brain many aspects. exploring how utilize competitive learning, adaptive regulation, synaptic plasticity optimize model's prediction results, adjust its parameters, achieve dynamic adjustments perspective neuroscience brain-like cognition is interest.Experimental results demonstrate that achieves outstanding performance recognition tasks within surpassing other existing models. This not only provides efficient accurate natural technology health field but also introduces new ideas directions design optimization
Язык: Английский
Процитировано
9Scientific Reports, Год журнала: 2024, Номер 14(1)
Опубликована: Окт. 28, 2024
Optimizing agricultural water resource management is crucial for food production, as effective can significantly improve irrigation efficiency and crop yields. Currently, precise demand forecasting have become key research focuses; however, existing methods often fail to capture complex spatial temporal dependencies. To address this, we propose a novel deep learning framework that combines remote sensing technology with the UNet-ConvLSTM (UCL) model effectively integrate features from MODIS GLDAS datasets. Our leverages high-resolution data UNet dependencies captured by ConvLSTM prediction accuracy. Experimental results demonstrate our UCL achieves best $$R^2$$ compared methods, reaching 0.927 on dataset 0.935 dataset. This approach highlights potential of AI technologies in addressing critical challenges management, contributing more sustainable efficient production systems.
Язык: Английский
Процитировано
3Alexandria Engineering Journal, Год журнала: 2025, Номер 116, С. 586 - 600
Опубликована: Янв. 8, 2025
Процитировано
0