Subspace learning has been widely applied for feature extraction of hyperspectral images (HSIs) and achieved great success. However, the current methods still leave two problems that need to be further investigated. Firstly, those methods mainly focus on finding one or multiple projection matrices for mapping the high-dimensional data into a low-dimensional subspace, which can only capture information from each direction of high-order hyperspectral data separately. Secondly, the feature extraction performance is barely satisfactory when HSI data is severely corrupted by noise. To address these issues, this paper presents a t-linear tensor subspace learning (tLTSL) model for robust feature extraction of HSIs based on t-product projection. In the model, t-product projection is a new defined tensor transformation way similar to linear transformation in vector space, which can maximally capture the intrinsic structure of tensor data. The integrated tensor low-rank and sparse decomposition can effectively remove the noise corruption and the learned t-product projection can directly transform the high-order HSI data into a subspace with information from all modes comprehensively considered. Moreover, a proposition related to tensor rank is proofed for interpreting the meaning of the tLTSL model. Extensive experiments are conducted on two different kinds of noise ( i.e ., simulated and real noise) corrupted HSI data, which validate the effectiveness of tLTSL.

This paper presents the analytical and numerical investigation on the global synchronization and anti-synchronization for a class of drive-response systems of fractional-order complex-valued gene regulatory networks with time-varying delays (DFGRNs). In our design, two kinds of adaptive feedback controllers are used to synchronize and anti-synchronize the proposed drive-response systems, and some sufficient conditions on the global asymptotical synchronization and anti-synchronization are given with the methods of the fractional Lyapunov-like functions and the fractional-order inequalities. In the numerical simulations, two minimum “estimated time", T1 and T2, are computed to achieve the synchronization and anti-synchronization.We find that T1 and T2 increase with the decreasing of the fractional order of DFGRNs.

To enhance the accessibility and accuracy of plant protection knowledge for agricultural practitioners, this study develops an intelligent question-answering (QA) system based on a large language model (LLM). A local knowledge base was constructed by vectorizing 7000 research papers and books in the field of plant protection, from which 568 representative papers were selected to generate QA data using an LLM. After data cleaning and filtering, a fine-tuning dataset comprising 9000 question–answer pairs was curated. To optimize the model’s performance, low-rank adaptation (LoRA) was applied to the InterLM-20B model, resulting in the InterLM-20B-LoRA, which was integrated with Langchain-ChatChat and the local knowledge base to develop the QA system. Additionally, retrieval-augmented generation (RAG) technology was implemented to enhance response accuracy by enabling the model to retrieve relevant field-specific knowledge before generating answers, effectively mitigating the risk of hallucinated information. The experimental results demonstrate that the proposed system achieves an answer accuracy of 97%, highlighting its potential as an advanced solution for intelligent agricultural QA services.

Background: Predicting the protein-ATP binding sites is a highly unbalanced binary classification problem, and higher precision prediction through the machine learning methods is of great significance to the researches on proteins’ functions and the design of drugs. Objective: Most existing researches typically select 17aa as the length of window by experience, and extract features by the Position-specific Scoring Matrix (PSSM), and then construct models predicting with SVC. However, the independent prediction values obtained in these researches are either over-high (ACC) or lower (MCC), and there is therefore a larger improvement room in the prediction precision. Methods: This paper utilizes the mutual information, I, to define the window length of 15aa, and the Pseudo Position Specific Scoring Matrix (PsePSSM), which is more fault-tolerance, to extract the features, and then train multiple 1:1 SVC classifiers to model, and finally perform the simple votings. Results: The prediction results over two protein-ATP binding site datasets, the ATP168 and the ATP227, are totally superior to the independent prediction results obtained in the Reference Feature Extraction Approach. And in our approach, the MCC values are respectively improved, from the range of 0.3110 ~ 0.5360 and the range of 0.3060 ~ 0.553, to 0.7512 and 0.7106. Conclusion: Further, we explain why the PsePSSM approach is more fault-tolerance. This approach has a promising application prospect in the feature-extraction of protein sequences.

The accurate measurement of soil organic matter (SOM) is vital for maintaining soil quality. We present an innovative model for SOM prediction by integrating spectral and profile features. We use PCA, Lasso, and SCARS methods to extract important spectral features and combine them with profile data. This hybrid approach significantly improves SOM prediction across various models, including Random Forest, ExtraTrees, and XGBoost, boosting the coefficient of determination (R²) by up to 26%. Notably, the ExtraTrees model, enriched with PCA-extracted features, achieves the highest accuracy with an R² of 0.931 and an RMSE of 0.068. Compared with single-feature models, this approach improves the R² by 17% and 26% for PCA features of full-band spectra and profile features, respectively. Our findings highlight the potential of feature integration, especially the ExtraTrees model with PCA-extracted features and profile features, as a stable and accurate tool for SOM prediction in extensive study areas.

Accurately obtaining both the number and the location of rice plants plays a critical role in agricultural applications, such as precision fertilization and yield prediction. With the rapid development of deep learning, numerous models for plant counting have been proposed. However, many of these models contain a large number of parameters, making them unsuitable for deployment in agricultural settings with limited computational resources. To address this challenge, we propose a novel pruning method, Cosine Norm Fusion (CNF), and a lightweight feature fusion technique, the Depth Attention Fusion Module (DAFM). Based on these innovations, we modify the existing P2PNet network to create P2P-CNF, a lightweight model for rice plant counting. The process begins with pruning the trained network using CNF, followed by the integration of our lightweight feature fusion module, DAFM. To validate the effectiveness of our method, we conducted experiments using rice datasets, including the RSC-UAV dataset, captured by UAV. The results demonstrate that our method achieves a MAE of 3.12 and an RMSE of 4.12 while utilizing only 33% of the original network parameters. We also evaluated our method on other plant counting datasets, and the results show that our method achieves a high counting accuracy while maintaining a lightweight architecture.