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. First, those mainly focus on finding one or multiple projection matrices mapping high-dimensional data into a low-dimensional subspace, which can only capture information from each direction high-order separately. Second, performance is barely satisfactory when severely corrupted...

Accurately obtaining both the number and location of rice plants plays a critical role in agricultural applications, such as precision fertilization yield prediction. With rapid development deep learning, numerous models for plant counting have been proposed. However, many these contain large parameters, making them unsuitable deployment settings with limited computational resources. To address this challenge, we propose novel pruning method, Cosine Norm Fusion (CNF), lightweight feature...

Barnyard grass, a pernicious weed thriving in rice fields, poses significant challenge to agricultural productivity. Detection of barnyard grass before the four-leaf stage is critical for effective control measures. However, due their striking visual similarity, separating them from seedlings at early growth stages daunting using traditional visible light imaging models. To explore feasibility hyperspectral identification and seedling stage, we have pioneered DeepBGS feature parsing...

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 base was constructed by vectorizing 7000 research papers books in field protection, from which 568 representative were selected to generate QA data using LLM. After cleaning filtering, fine-tuning dataset comprising 9000 question–answer pairs curated. optimize model’s...

Abstract Although low‐rank representation (LRR)‐based subspace learning has been widely applied for feature extraction in computer vision, how to enhance the discriminability of low‐dimensional features extracted by LRR based methods is still a problem that needs be further investigated. Therefore, this paper proposes novel preserving embedding regression (LRPER) method integrating LRR, linear regression, and projection into unified framework. In LRPER, can reveal underlying structure...

Cytoplasmic male sterility (CMS) is a complex phenomenon of plant that can produce non-functional pollen. It caused by mutation, rearrangement or recombination in the mitochondrial genome. So far, systematic structural characteristics changes genome from maintainer lines to CMS have not been reported tobacco.The genomes flower buds both and two Nicotiana tabacum cultivars (YY85, sYY85, ZY90, sZY90) were sequenced using PacBio Illumina Hiseq technology, several findings produced comparative...

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

Hyperspectral images (HSIs) capture a wide range of spectral features across multiple bands light, from visible to near-infrared. image classification technology enables researchers accurately identify and analyze the composition distribution surface materials. Current mainstream deep learning methods typically use block sampling spatial for model. However, this approach can affect results due influence neighboring within sample block. To improve model’s focus on center block, study proposes...

Since the decision trees (DTs) have an advantage over "black-box" models, such as neural nets or support vector machines, in terms of comprehensibility, that it might merit improvement for further optimization. The node splitting measures and pruning methods are primary among techniques can improve generalization abilities DTs. Here, we introduced unequal interval optimization splitting, well local chi-square test tree pruning. This new method was named adaptive multi-branch (CMDT). 11...

The environmental protection agency thinks that quantitative structure–activity relationship (QSAR) analysis can better replace toxicity tests.

For precision medicine, there is a need to identify genes that accurately distinguish the physiological state or response particular therapy, but this can be challenging. Many methods of analyzing differential expression have been established and applied problem, such as t-test, edgeR, DEseq2. A common feature these their focus on linear relationship (differential expression) between gene phenotype. However, they may overlook nonlinear relationships due various factors, degree disease...

Insect recognition, crucial for agriculture and ecology studies, benefits from advancements in RGB image-based deep learning, yet still confronts accuracy challenges. To address this gap, the HI30 dataset is introduced, comprising 2115 hyperspectral images across 30 insect categories, which offers richer information than data enhancing classification accuracy. effectively harness dataset, study presents Two-Branch Self-Correlation Network (TBSCN), a novel approach that combines spectrum...

(1) Background: The striped rice stem borer (SRSB), Chilo suppressalis, has severely diminished the yield and quality of in China. A timely accurate prediction pest population can facilitate designation a control strategy. (2) Methods: In this study, we applied multiple linear regression (MLR), gradient boosting decision tree (GBDT), deep auto-regressive (DeepAR) models dynamic SRSB occurrence during crop season from 2000 to 2020 Hunan province, China, by using weather factors time series...

The accurate measurement of soil organic matter (SOM) is vital for maintaining quality. We present an innovative model SOM prediction by integrating spectral and profile features. use PCA, Lasso, SCARS methods to extract important features combine them with data. This hybrid approach significantly improves across various models, including Random Forest, ExtraTrees, XGBoost, boosting the coefficient determination (R2) up 26%. Notably, ExtraTrees model, enriched PCA-extracted features,...

Cross-modal hashing is widely used in multimedia retrieval tasks for its advantages of low storage cost and high speed. In recent years, many deep unsupervised cross-modal methods were proposed to deal with unlabeled data. These usually generate a similarity matrix from raw features as guidance signal model training. However, this type techniques introduce bias feature measurements they fail accurately capture semantic relationships between data eliminate the error caused by original...

Abstract Considering the issue with respect to high data redundancy and cost of information collection in wireless sensor nodes, this paper proposes a fusion method based on belief structure reduce attribution multi-granulation rough set. By introducing structure, attribute reduction is carried out for sets. From view granular computing, studies evidential characteristics incomplete ordered systems. On basis, positive region reduction, plausibility are put forward system analyze consistency...

Abstract Background Sphingobacterium is a class of Gram-negative, non-fermentative bacilii, and rarely involved in human infections. It characterized by large number cellular membrane sphingophospolipids. Due to its wide ecological distribution oil degradation ability, environmental microbiologists have paid much attention it. Results A novel gram-negative bacterium, designated CZ-2 T , was isolated from sample tobacco leaves infected with wildfire disease Guiyang County, Chenzhou City,...

Cytosine (C) to uracil (U) RNA editing is one of the most important post-transcriptional processes, however exploring C-to-U events efficiently within crop mitochondrial genome remains a challenge. An improving predictive editor for genomes, iPReditor-CMG, was proposed, which based on SVM, three common genomes and self-sequenced tobacco ATPase. After multi-combination feature extracting, high-dimension screening multi-test independent predicting, results showed that average accuracy...