The COVID-19 pandemic poses a serious threat to global public health. Anti-coronavirus peptides (ACVPs) exhibit high targeting, low toxicity, and excellent modifiability, making them promising candidates for antiviral drug discovery. These properties offer advantages over traditional small-molecule drugs. To reduce the time and cost spent on large-scale peptide screening and activity validation, there is an urgent need to construct efficient artificial intelligence models that assist in the identification of ACVPs. However, the limited number of experimentally validated ACVPs severely constrains the generalization ability and prediction accuracy of existing computational models. In this study, a new prediction framework, the Multi-modal Feature Extraction and Ensemble learning framework for Anti-Coronavirus Peptide prediction(MFE-ACVP), is proposed for identifying potential candidate peptides for ACVPs. The method generates high-quality ACVPs by introducing an improved Generative Adversarial Network (GAN) with materialization constraints to alleviate the problem of data insufficiency. Simultaneously, fusing sequence, structural, evolutionary, and topological features, we constructed a 100-dimensional cross-scale feature representation. An ensemble architecture integrating five traditional machine learning models with deep neural networks (DNNs) was designed to enhance predictive performance. Compared with the existing models PreAntiCoV, iACVP, ACVPred, and ENNAVIA-C/D, MFE-ACVP achieved 86.37%, 77.62%, and 65.19% of area under the curve (AUC), accuracy (ACC), and Matthew’s correlation coefficient (MCC), respectively, on an independent validation set, showing superior predictive performance and stability. To facilitate ACVP screening, we developed a publicly accessible web server http://bioprediction.sa1.tunnelfrp.com/.

Background: Bovine viral diarrhea virus (BVDV) can cause diarrhea, abortion, and immunosuppression in cattle, imposing huge economic losses for the global cattle industry. The pathogenic and immune mechanisms of BVDV remain elusive. The development of a BVDV-gene knowledge base can provide clues to reveal the interaction of BVDV with host cells. However, the traditional method of manually establishing a knowledge base is time-consuming and inefficient. The method of developing a knowledge base based on deep learning has noticeably attracted scholars' attention recently. Objective: The study aimed to explore the substitution of deep learning for manual mining of BVDVrelated genes and to develop a knowledge graph of the relationship between BVDV and related genes. Methods: A deep learning-based biomedical knowledge graph development method was proposed, which used deep learning to mine biomedical knowledge, model BVDV and various gene concepts, and store data in a graphical database. First, the PubMed database was used as the data source and crawler technology to obtain abstract data on the relationship between BVDV and various host genes. Pretrained BioBERT model was used for biomedical named entity recognition to obtain all types of gene entities, and the pre-trained BERT model was utilized for relationship extraction to achieve the relationship between BVDV and various gene entities. Then, it was combined with manual proofreading to obtain structured triple data with high accuracy. Finally, the Neo4j graph database was used to store data and to develop the knowledge graph of the relationship between BVDV and related genes. Results: The results showed the obtainment of 71 gene entity types, including PRL4, MMP-7, TGIF1, etc. 9 relation types of BVDV and gene entities were obtained, including "can downregulate expression of", "can upregulate expression of", "can suppress expression of", etc. The knowledge graph was developed using deep learning to mine biomedical knowledge combined with manual proofreading, which was faster and more efficient than the traditional method of establishing knowledge base manually, and the retrieval of semantic information by storing data in graph database was also more efficient. Conclusion: A BVDV-gene knowledge graph was preliminarily developed, which provided a basis for studying the interaction between BVDV and host cells.

Aiming at the differences in spatial spectral attributes between forested land and cultivated land on remote sensing images, and the deficiencies of existing remote sensing change detection methods that are difficult to capture fine edge structures and distinguish pseudo changes, this paper introduces an Edge Enhanced Hierarchical Dual Gated Change Detection (EHDCD) model for forested land and cultivated land, aiming to meet the demand for representing the complex features of these two land types. The model designs an Edge Enhanced Channel Attention Module (EECA) to strengthen the edge recognition ability and suppress the noise interference; proposes a High-Low Level Dynamic Adaptation Strategy (HiLo) to realize the balanced expression of detail information and semantic features; and constructs a Dual Gated Feature Compensation Module (DGFM) to effectively reduce the misdetection rate of change detection. Experiments show that the F1 scores of the model on the self-constructed forest and agricultural dataset FC-CD and public datasets CLCD and SYSU-CD reach 89.06%, 83.37%, and 85.06%, respectively, which can more accurately support the dynamic monitoring applications of forest land and cropland.

The joint extraction of entities and relations is a critical task in information extraction, and its performance directly affects the performance of downstream tasks. However, existing joint extraction models based on deep learning exhibit weak processing capabilities for the phenomenon of multiple pronunciations of one character and multiple characters of one pronunciation when processing Chinese texts, resulting in a performance loss. To address these issues, this paper introduces part-of-speech (POS) and pinyin features to aid the model in learning semantic features that are more contextually appropriate. We propose a Chinese Joint Entity and Relation Extraction Model based on progressive contrastive learning and heterogeneous feature fusion (CJE-PCHF). During model training, an interactive fusion network based on progressive contrastive learning is employed to learn the dependencies between pinyin, POS, and semantic features. This guides the model in heterogeneous feature fusion, capturing higher-order semantic associations between heterogeneous features. On the commonly used DuIE evaluation dataset for joint extraction, our model achieved a significant improvement, with the F1 score increasing by 5.4% compared to the benchmark model CasRel.

Accurate crop classification plays a critical role in agricultural monitoring and food security assurance. Effectively exploiting spatiotemporal information from multi-temporal remote sensing data remains a key challenge in crop mapping.This study proposes an improved neural network model, termed the (3+2)D Split-Attention Feature Pyramid Network ((3+2)D SAFPN), which is built upon a hybrid 3D–2D Feature Pyramid Network ((3+2)D FPN). The model integrates a 3D FPN to capture spatiotemporal crop dynamics, a 2D FPN to extract multi-scale spatial features, a split-attention (SA) mechanism to enhance inter-channel information interaction, and a focal loss function to improve learning performance on minority crop classes. Multi-temporal Sentinel-2 imagery acquired in 2024 was used to construct a plot-level NDVI time-series dataset for Talhu Town, Wuyuan County, Bayannur City, Inner Mongolia. The dataset was divided into training, validation, and test sets with a ratio of 6:2:2.Experimental results demonstrate that the proposed (3+2)D SAFPN model achieved overall accuracies of 89.01% and 89.06% on the test and validation sets, respectively, with Kappa coefficients of 0.82 for both sets, outperforming the original (3+2)D FPN model. Furthermore, comparative experiments conducted on the public Munich dataset indicate strong generalization ability, with accuracy improvements of 2.88% on the test set and 2.44% on the validation set compared to the baseline model.The results indicate that the (3+2)D SAFPN model effectively integrates spatial, spectral, and temporal information from multi-temporal remote sensing imagery, providing a robust and high-accuracy solution for crop classification tasks. This approach shows strong potential for large-scale agricultural monitoring applications. The source code of the proposed model is publicly available at: https://gitee.com/btgw/YicongSun/ree/(3+2)D-SAFPN_torch .