Search

The construction of hypernym taxonomic trees, a critical task in the field of natural language processing, involves extracting lexical relationships, specifically creating a tree structure that represents hypernym relationships among a given set of words within the same domain. In this work, we present a method for constructing hypernym taxonomy trees in the Chinese language domain, and we named it CHRRM (Chinese Hypernym Relationship Reasoning Model). Our method consists of two main steps: First, we utilize pre-trained models to predict hypernym relationships between pairs of words; second, we regard these relationships as edges to form a maximum spanning tree in the word graph. Our method enhances the effectiveness of constructing hypernym taxonomic trees based on pre-trained models through two key improvements: (1) We optimize the hyperparameter configuration for this task using pre-trained models from the Bert family and provide explanations for the configuration of these hyperparameters. (2) By employing generative large language models such as ChatGPT and ChatGLM to annotate words, we improve the accuracy of hypernym relationship identification and analyze the feasibility of applying generative large language models to the task of constructing taxonomy trees. We trained our model on subtrees of WORDNET and evaluated its performance on non-overlapping subtrees of WORDNET, demonstrating that our enhancements led to a significant relative improvement of 15.67%, achieving an F1 score of 67.9 on the Chinese WORDNET validation dataset compared to the previous score of 58.7. In conclusion, our study reveals the following key findings: (1) The Roberta-wwm-ext-large model consistently delivers outstanding results in constructing taxonomic trees. (2) Generative large language models, while capable of aiding pre-trained models in improving hypernym recognition accuracy, have limitations related to generation quality and computational resources. (3) Generative large language models can serve various NLP tasks either directly or indirectly; it is feasible to improve the downstream NLU task’s performance through the generative content.

Abstract Proper pericarp thickness protects the maize kernel against pests and diseases, moreover, thinner pericarp improves the eating quality in fresh corn. In this study, we aimed to investigate the dynamic changes in maize pericarp during kernel development and identified the major quantitative trait loci (QTLs) for maize pericarp thickness. It was observed that maize pericarp thickness first increased and then decreased. During the growth and formation stages, the pericarp thickness gradually increased and reached the maximum, after which it gradually decreased and reached the minimum during maturity. To identify the QTLs for pericarp thickness, a BC4F4 population was constructed using maize inbred lines B73 (recurrent parent with thick pericarp) and Baimaya (donor parent with thin pericarp). In addition, a high-density genetic map was constructed using maize 10 K SNP microarray. A total of 17 QTLs related to pericarp thickness were identified in combination with the phenotypic data. The results revealed that the heritability of the thickness of upper germinal side of pericarp (UG) was 0.63. The major QTL controlling UG was qPT1-1, which was located on chromosome 1 (212,215,145–212,948,882). The heritability of the thickness of upper abgerminal side of pericarp (UA) was 0.70. The major QTL controlling UA was qPT2-1, which was located on chromosome 2 (2,550,197–14,732,993). In addition, a combination of functional annotation, DNA sequencing analysis and quantitative real-time PCR (qPCR) screened two candidate genes, Zm00001d001964 and Zm00001d002283, that could potentially control maize pericarp thickness. This study provides valuable insights into the improvement of maize pericarp thickness during breeding.

Abstract Zero-shot learning represents a formidable paradigm in machine learning, wherein the crux lies in distilling and generalizing knowledge from observed classes to novel ones. The objective is to identify unfamiliar objects that were not included in the model’s training, leveraging learned patterns and knowledge from previously encountered categories. As a crucial subtask of open-world object detection, zero-shot classification can also provide insights and solutions for this field. Despite its potential, current zero-shot classification models often suffer from a performance gap due to limited transfer ability and discriminative capability of learned representations. In pursuit of advancing the subpar state of zero-shot object classification, this paper introduces a novel model for image classification which can be applied to object detection, namely, self-distillation and k-nearest neighbor-based zero-shot classification method. First, we employ a diffusion detector to identify potential objects in images. Then, self-distillation and distance-based classifiers are used for distinguishing unseen objects from seen classes. The k-nearest neighbor-based cluster heads are designed to cluster the unseen objects. Extensive experiments and visualizations were conducted on publicly available datasets on the efficacy of the proposed approach. Precisely, our model demonstrates performance improvement of over 20% compared to contrastive clustering. Moreover, it achieves a precision of 0.910 and a recall of 0.842 on CIFAR-10 datasets, a precision of 0.737, and a recall of 0.688 on CIFAR-100 datasets for the macro average. Compared to a more recent model (SGFR), our model realized improvements of 10.9%, 13.3%, and 7.8% in Sacc, Uacc, and H metrics, respectively. This study aims to introduce fresh ideas into the domain of zero-shot image classification, and it can be applied to open-world object detection tasks. Our code is available at https://www.github.com/CmosWolf1/Code_implementation_for_paper_SKZC .

Cloud is a serious problem that affects the quality of remote-sensing (RS) images. Existing cloud removal techniques suffer from notable limitations, such as being specific to certain data types, cloud conditions, and spatial extents, as well as requiring auxiliary data, which hampers their generalizability and flexibility. To address the issue, we propose a maximum-value compositing approach by generating cloud masks. We acquired 432 daily MOD09GA L2 MODIS imageries covering a vast region with persistent cloud cover and various climates and land-cover types. Labeled datasets for cloud, land, and no-data were collected from selected daily imageries. Subsequently, we trained and evaluated RF, SVM, and U-Net models to choose the best models. Accordingly, SVM and U-Net were chosen and employed to classify all the daily imageries. Then, the classified imageries were converted to two sets of mask layers to mask clouds and no-data pixels in the corresponding daily images by setting the masked pixels’ values to −0.999999. After masking, we employed the maximum-value technique to generate two sets of 16-day composite products, MaxComp-1 and MaxComp-2, corresponding to SVM and U-Net-derived cloud masks, respectively. Finally, we assessed the quality of our composite products by comparing them with the reference MOD13A1 16-day composite product. Based on the land-cover classification accuracy, our products yielded a significantly higher accuracy (5–28%) than the reference MODIS product across three classifiers (RF, SVM, and U-Net), indicating the quality of our products and the effectiveness of our techniques. In particular, MaxComp-1 yielded the best results, which further implies the superiority of SVM for cloud masking. In addition, our products appear to be more radiometrically and spectrally consistent and less noisy than MOD13A1, implying that our approach is more efficient in removing shadows and noises/artifacts. Our method yields high-quality products that are vital for investigating large regions with persistent clouds and studies requiring time-series data. Moreover, the proposed techniques can be adopted for higher-resolution RS imageries, regardless of the spatial extent, data volume, and type of clouds.

Abstract Intraoperative ingestion of blood, cerebrospinal fluid, and irrigation fluid can lead to an increase in gastric volume, resulting in the potential risk of aspiration in patients after endoscopic endonasal transsphenoidal surgery (EETS). In this prospective observational study, we aimed to assess the volume of gastric contents in patients undergoing this neurosurgical procedure using ultrasound, and to determine the factors associated with volume change. Eighty-two patients diagnosed with pituitary adenoma were recruited consecutively. Semi-quantitative (Perlas scores: 0, 1 and 2) and quantitative (cross-sectional area, CSA) ultrasound assessments of the gastric antrum were performed immediately before and after surgery in the semi-recumbent and semi-recumbent right-lateral positions. Seven (8.5%) patients had antrum scores from preoperative grade 0 to postoperative grade 2; nine (11%) patients had antrum scores from preoperative grade 0 to postoperative grade 1. The mean ± standard deviation (SD) of increased gastric volume was 71.0 ± 33.1 mL and 236.5 ± 32.4 mL in postoperative grade 1 and 2 groups, respectively. Subgroup analysis showed that 11 (13.4%) patients (4 in grade 1 and all in grade 2) had postoperative estimated gastric volume > 1.5 mL kg−1 (mean ± SD 3.08 ± 1.67, range 1.51–5.01 mL kg−1). Logistic regression analysis revealed that older age, diabetes mellitus, and long surgical duration were independent risk factors for significant volume change (all P

A method was developed for the determination of 14 anabolic androgenic steroids including trenbolone, boldenone and nandrolone in animal derived foods by liquid chromatography-tandem mass spectrometry. The samples were added to ammonium acetate buffer and enzymatic hydrolyzed by β-glucuronidase and arysulfatase. The hydrolyzates were extracted with 20 mL acetonitrile, and then 10 mL supernatant was degreased by 10 mL acetonitrile saturated hexane, purified with 50 mg C18, 50 mg PSA and 300 mg neutral alumina powder including 300 mg MgSO4. Then the analytes were separated by C18 column with gradient elution using 0.1% formic acid and acetonitrile as the mobile phase, detected in positive multiple reaction monitoring (MRM) and quantified with matrix-matched external standards. The linear regression correlation coefficients for the compounds were all higher than 0.992 in the range from 0.1 to 10 ng/mL. The limits of detection and the limits of quantification were 0.3 and 1.0 μg/kg respectively. The recoveries at the addition levels of 1.0, 2.0 and 10 μg/kg were in the range of 75.7%~104.1%, with relative standard deviations (RSDs) of 3.60%~12.1%. This method had high effectivity, good reproducibility, simple operation and high efficiency, and could be used for the rapid determination needs of common anabolic androgenic steroids in foodborne stimulants in the Winter Olympic Games.

Manufacturing errors are one of the main causes of vibration and noise of gear transmission system，and the combination of multiple errors seriously affects the meshing performance of gear pairs. The influence degree and law of various manufacturing errors on meshing performance of herringbone gear pairs are mainly studied. Firstly，the manufacturing error model is established，and the actual tooth surface is constructed based on the theory of spatial coordinate transformation. Secondly，the influence law of different errors on the meshing performance of the gear pair is explored through the load distribution on the tooth surface，and the influence degree of different errors is explored based on the orthogonal experiment. Finally，the relationship between the influence of typical error types on the meshing performance of gear pairs and the system load is analyzed. The results show that the influence law of various errors on the meshing performance of gear pair is different，and the positive tilt deviation of helix direction has the most prominent influence on the meshing performance of gear pairs under heavy load.

An accurate calculation method of the maximum grinding wheel diameter for the form grinding of modification herringbone gears is proposed. A modified surface with three-section parabolic tooth profile is established. According to the meshing equation of the grinding wheel and the workpiece，the contact line equation of the grinding wheel is derived. The contact line of the grinding wheel is moved spirally around the axis of the workpiece，and considering the change of the center distance of the grinding wheel，a double modification of the herringbone gear is realized. According to the geometric conditions，when the grinding wheel is grinding the end point of the tooth surface，the maximum diameter of the grinding wheel is calculated，and the influence of the basic parameters，modification parameters and process parameters of the gear on the maximum diameter of the grinding wheel is analyzed. Numerical example simulation shows that tooth surface modification has little effect on the maximum diameter of the grinding wheel，and grinding depths have a greater influence on the maximum diameter of the grinding wheel. Both the number of teeth and the helix angle have an obvious effect on the maximum diameter of the grinding wheel.

Considerable genetic variation exists in maize ear size and kernel number. Here the authors show that variation in a gene encoding an ethylene biosynthetic enzyme impacts ear length, flower fertility and kernel yield suggesting an important role for ethylene signaling during inflorescence development.

Salt stress hinders plant growth and disrupts many physio-biochemical and molecular processes. The present study evaluated the potential role of foliar applied 24-epibrassinolide (EBR) for mitigating the adverse effects of salinity in rice plants. Our findings showed that salt stress reduced several growth and physiological parameters, and substantially increased oxidative stress in rice seedlings. The foliar application of EBR replenished the negative effects of salt stress by improving plant growth and physiological attributes in rice plants. Moreover, application of EBR substantially reduced oxidative injuries by enhancing antioxidant enzymes and phenolic compounds in rice plants during salt stress. Interestingly, the endogenous ABA levels were declined in stressed plants by EBR treatment, also validated by expression analysis of ABA-related genes. In addition, application of EBR upregulated carotenoids and flavonoids pathway genes, thus showing its function to activate secondary metabolism in rice plants during salt stress. The principal component analysis (PCA) indicated that, under salt stress, the response of EBR-treated rice plants was strongly correlated with the accumulation of phenolic compounds and flavonoids. Collectively, this study presents useful insight into the distinct role of EBR by activating underlying enzymatic and non-enzymatic defense mechanisms to mitigate salinity in rice plants.

Abstract Background The objective of this study was to investigate genetic variations and the relationships between these genetic variations and clinicopathological features of high-recurrence risk papillary thyroid carcinoma in a southern Chinese population. Methods One hundred sixty-eight patients of high-recurrence risk papillary thyroid carcinoma were recruited for this study from 2017 to 2018. Formalin-fixed paraffin-embedded tissue and the data of clinicopathological characteristics were all collected and analyzed from these patients. We used next-generation sequencing technology to investigate the targeted gene mutations and gene fusions of the pathology specimens. Results The frequency of candidate tumor driver gene mutation was 85.1% in 143 patients, including BRAF V600E mutation in 119 patients(70.8%), RET fusion in 13 patients(7.7%), TERT promoter mutations in 11 patients(6.5%), RAS (HRAS, NRAS, KRAS) gene mutations in 10 patients(6.0%), and other mutations involving TP53, PIK3CA, AKT1, PTEN and NTRK1. Concomitant presence of more than two genetic aberrations was seen in 27 patients (16.1%). Our study showed that BRAF V600E mutation is highly correlated with conventional PTC (p 45, p = 0.003) and higher disease stage of III or IV (p = 0.002). RAS gene and BRAF V600E co-mutations were only seen in multifocal PTC (p = 0.015). Conclusion In our high-recurrence risk PTC cohort, most patients had more than one driver gene aberration. Coexistence of BRAF V600E with TERT promoter mutations or with RAS mutations were significantly correlated with worse clinicopathological characteristics.

Selection of kernel number per ear has improved maize yield, but the genetic base is unclear. Here, the authors reveal that a serine/threonine protein kinase KNR6 is a positive regulator of the trait and show in vitro evidences that KNR6 may function through phosphorylating an Arf GTPase-activating protein.

A new dynamic model for a two-input two-path split torque transmission system which considers meshing error, time-varying meshing stiffness, and meshing-in impact is proposed. Time-varying meshing stiffness and meshing-in impact of each gear pair are accurately calculated based on tooth contact analysis and loaded tooth contact analysis. Equivalent displacements of eccentricity error and installation error along the meshing line of second- and third-stages gears are derived. The modified tooth surface of a third-stage double-helical gear is obtained by optimizing the amplitude of static loaded transmission error and meshing-in impact via nondominated sorting genetic algorithm-II (NSGA-II). Influence of modification on load sharing and dynamic load characteristics of split torque transmission system is investigated. The results indicate that the system’s dynamic meshing force increases when meshing-in impact is accounted for, which is unfavorable for the transmission. Following the modification of a double-helical gear, the dynamic load characteristics of the split torque transmission system are significantly improved, while its load sharing characteristics are improved to a certain extent.