Large language models(LLMs)show great potential in educational scenarios but face challenges like hallucination,knowledge gaps,and reasoning discontinuities.This study proposes a dynamic knowledge enhancement framewor...Large language models(LLMs)show great potential in educational scenarios but face challenges like hallucination,knowledge gaps,and reasoning discontinuities.This study proposes a dynamic knowledge enhancement framework.By integrating local knowledge graphs and stepwise prompting mechanisms,it improves LLMs’accuracy and interpretability in solving professional domain problems.The framework has two core modules:an LLM-driven knowledge graph construction system for incremental updates and a unified reasoning engine for generating enhanced prompts.Experiments on 680 educational questions show that the method boosts accuracy by 4.5%and 4.3%for multi-step reasoning and knowledge-dependent questions respectively,and increases reasoning step completeness from 68.2%to 83.7%.It also reduces hallucination problems.Key contributions include the followings:①validation of an effective framework synergizing knowledge graphs with retrieval mechanisms to enhance LLM reliability;②a stepwise prompting strategy enforcing explicit reasoning chain generation,addressing pedagogical requirements for process interpretability;③a lightweight deployment solution for educational systems such as adaptive learning platforms.展开更多
Nest predation is the leading cause of reproductive failure in birds and a major driving force in the evolution of anti-predation strategies.Current studies on the anti-predation strategies of birds driven by predatio...Nest predation is the leading cause of reproductive failure in birds and a major driving force in the evolution of anti-predation strategies.Current studies on the anti-predation strategies of birds driven by predation pressure have mainly focused on adults.However,the detection and behavioral responses of nestlings toward predation risk require further investigation.In this study,we examined nestling responses to predator sounds.Two species of nestlings,the Reed Parrotbill(Paradoxornis heudei)and Vinous-throated Parrotbill(Sinosuthora webbiana),were exposed to the Sparrowhawk(Accipiter nisus,less common)and Oriental Magpie(Pica serica,more common),which are predator species,the Oriental Turtle Dove(Streptopelia orientalis),which is a commonly found harmless species,and background noise.Our findings revealed that compared to pre-playback of natural begging and playback of background noise and Oriental Turtle Dove sounds,playback of the two predator types decreased the total begging time and total number of calls of the two nestlings species,with the calls of the Sparrowhawk leading to greater suppression of nestling begging behavior than those of the Oriental Magpie.Therefore,our results indicated that these nestlings were able to distinguish predators from harmless species based on auditory cues.展开更多
Objectives:Progression to castration-resistant prostate cancer(CRPC)and metastasis are the greatest challenges to effective treatment.Anticancer strategies targeting the key kinases associated with the development of ...Objectives:Progression to castration-resistant prostate cancer(CRPC)and metastasis are the greatest challenges to effective treatment.Anticancer strategies targeting the key kinases associated with the development of CRPC may represent a breakthrough.The tyrosine kinase receptor Erythropoietin-producing hepatocellular(Eph)A2 receptor is highly expressed in CRPC cell lines and may be associated with tumor invasion and metastasis.However,the effects and exact mechanisms of EphA2 in CRPC are only partially understood.This study aimed to investigate the impact of EphA2 on CRPC cell behaviors and underlyingmolecular pathways.Methods:CRISPR/Cas9-mediated gene editing induced EphA2-disrupted in human-derived PC3 andDU145 cells.Single-guideRNAs(sgRNAs)targeting EphA2 were designed,and editing efficiency was validated.Optimal sgRNA sequences were selected to generate EphA2-knockdown(KD)and-overexpressing(OE)cell lines.Cell migration,proliferation,and apoptosis were assessed via functional assays.Transcriptomic analysis,quantitative PCR,and Western blotting were performed to identify downstream effectors.Bioinformatics analyses were used to correlate EphA2 and CDH1 expression with clinical parameters in prostate cancer patients.Results:Editing efficiency was found to vary among different sgRNAs targeting the EphA2 gene.EphA2-KD significantly inhibited CRPC cell migration but did not affect cell proliferation or apoptosis.Conversely,EphA2-OE significantly enhanced the migration of DU145 cells.Molecular analyses revealed that the expression of CDH1(an important marker of the epithelial-mesenchymal transition(EMT)in tumors)was significantly upregulated in PC3-EphA2-KD cells and downregulated in DU145-EphA2-OE cells,indicating that CDH1 is a downstream regulator of EphA2.Bioinformatic analysis revealed that higher EphA2 levels and lower CDH1 expression were both associated with an advanced tumor T stage,higher Gleason scores,and lymph nodemetastases in prostate cancer patients.More importantly,EphA2 was found to be an important predictor of lymph nodemetastasis,in addition to the Gleason score.Adding EphA2 to the Gleason score could significantly improve the detection of lymph node metastasis.Conclusion:CRISPR/Cas9-mediated EphA2-KD significantly suppressed the migration of CRPC cells through the inhibition of the EphA2-CDH1 axis.Strategies targeting the EphA2 genemay be promising for the treatment of CRPC.展开更多
Recent advances in convolution neural network (CNN) have fostered the progress in object recognition and semantic segmentation, which in turn has improved the performance of hyperspectral image (HSI) classification. N...Recent advances in convolution neural network (CNN) have fostered the progress in object recognition and semantic segmentation, which in turn has improved the performance of hyperspectral image (HSI) classification. Nevertheless, the difficulty of high dimensional feature extraction and the shortage of small training samples seriously hinder the future development of HSI classification. In this paper, we propose a novel algorithm for HSI classification based on three-dimensional (3D) CNN and a feature pyramid network (FPN), called 3D-FPN. The framework contains a principle component analysis, a feature extraction structure and a logistic regression. Specifically, the FPN built with 3D convolutions not only retains the advantages of 3D convolution to fully extract the spectral-spatial feature maps, but also concentrates on more detailed information and performs multi-scale feature fusion. This method avoids the excessive complexity of the model and is suitable for small sample hyperspectral classification with varying categories and spatial resolutions. In order to test the performance of our proposed 3D-FPN method, rigorous experimental analysis was performed on three public hyperspectral data sets and hyperspectral data of GF-5 satellite. Quantitative and qualitative results indicated that our proposed method attained the best performance among other current state-of-the-art end-to-end deep learning-based methods.展开更多
基金supported in part by the China-Singapore International Joint Research Institute(CSIJRI)under Grant No.206-A023001the Undergraduate Teaching Reform Project of Shandong University under Grant Nos.2023Y235 and 2025Y99.
文摘Large language models(LLMs)show great potential in educational scenarios but face challenges like hallucination,knowledge gaps,and reasoning discontinuities.This study proposes a dynamic knowledge enhancement framework.By integrating local knowledge graphs and stepwise prompting mechanisms,it improves LLMs’accuracy and interpretability in solving professional domain problems.The framework has two core modules:an LLM-driven knowledge graph construction system for incremental updates and a unified reasoning engine for generating enhanced prompts.Experiments on 680 educational questions show that the method boosts accuracy by 4.5%and 4.3%for multi-step reasoning and knowledge-dependent questions respectively,and increases reasoning step completeness from 68.2%to 83.7%.It also reduces hallucination problems.Key contributions include the followings:①validation of an effective framework synergizing knowledge graphs with retrieval mechanisms to enhance LLM reliability;②a stepwise prompting strategy enforcing explicit reasoning chain generation,addressing pedagogical requirements for process interpretability;③a lightweight deployment solution for educational systems such as adaptive learning platforms.
基金funded by the National Natural Science Foundation of China (32301295 to J.W.,32101242 to L.M.and 32260253 to L.W.)High-Level Talents Research Start-Up Project of Hebei University (521100222044 to J.W.)。
文摘Nest predation is the leading cause of reproductive failure in birds and a major driving force in the evolution of anti-predation strategies.Current studies on the anti-predation strategies of birds driven by predation pressure have mainly focused on adults.However,the detection and behavioral responses of nestlings toward predation risk require further investigation.In this study,we examined nestling responses to predator sounds.Two species of nestlings,the Reed Parrotbill(Paradoxornis heudei)and Vinous-throated Parrotbill(Sinosuthora webbiana),were exposed to the Sparrowhawk(Accipiter nisus,less common)and Oriental Magpie(Pica serica,more common),which are predator species,the Oriental Turtle Dove(Streptopelia orientalis),which is a commonly found harmless species,and background noise.Our findings revealed that compared to pre-playback of natural begging and playback of background noise and Oriental Turtle Dove sounds,playback of the two predator types decreased the total begging time and total number of calls of the two nestlings species,with the calls of the Sparrowhawk leading to greater suppression of nestling begging behavior than those of the Oriental Magpie.Therefore,our results indicated that these nestlings were able to distinguish predators from harmless species based on auditory cues.
基金funded by the National Natural Science Foundation of China(Grant No.81801754)the Suzhou Science and Technology Bureau Development Plan(Grant No.SYS2020147)the 26th Batch of Extra-Curricular Academic Research Fund Project of Soochow University(Grant No.KY2024235B).
文摘Objectives:Progression to castration-resistant prostate cancer(CRPC)and metastasis are the greatest challenges to effective treatment.Anticancer strategies targeting the key kinases associated with the development of CRPC may represent a breakthrough.The tyrosine kinase receptor Erythropoietin-producing hepatocellular(Eph)A2 receptor is highly expressed in CRPC cell lines and may be associated with tumor invasion and metastasis.However,the effects and exact mechanisms of EphA2 in CRPC are only partially understood.This study aimed to investigate the impact of EphA2 on CRPC cell behaviors and underlyingmolecular pathways.Methods:CRISPR/Cas9-mediated gene editing induced EphA2-disrupted in human-derived PC3 andDU145 cells.Single-guideRNAs(sgRNAs)targeting EphA2 were designed,and editing efficiency was validated.Optimal sgRNA sequences were selected to generate EphA2-knockdown(KD)and-overexpressing(OE)cell lines.Cell migration,proliferation,and apoptosis were assessed via functional assays.Transcriptomic analysis,quantitative PCR,and Western blotting were performed to identify downstream effectors.Bioinformatics analyses were used to correlate EphA2 and CDH1 expression with clinical parameters in prostate cancer patients.Results:Editing efficiency was found to vary among different sgRNAs targeting the EphA2 gene.EphA2-KD significantly inhibited CRPC cell migration but did not affect cell proliferation or apoptosis.Conversely,EphA2-OE significantly enhanced the migration of DU145 cells.Molecular analyses revealed that the expression of CDH1(an important marker of the epithelial-mesenchymal transition(EMT)in tumors)was significantly upregulated in PC3-EphA2-KD cells and downregulated in DU145-EphA2-OE cells,indicating that CDH1 is a downstream regulator of EphA2.Bioinformatic analysis revealed that higher EphA2 levels and lower CDH1 expression were both associated with an advanced tumor T stage,higher Gleason scores,and lymph nodemetastases in prostate cancer patients.More importantly,EphA2 was found to be an important predictor of lymph nodemetastasis,in addition to the Gleason score.Adding EphA2 to the Gleason score could significantly improve the detection of lymph node metastasis.Conclusion:CRISPR/Cas9-mediated EphA2-KD significantly suppressed the migration of CRPC cells through the inhibition of the EphA2-CDH1 axis.Strategies targeting the EphA2 genemay be promising for the treatment of CRPC.
基金the National Natural Science Foundation of China(No.51975374)。
文摘Recent advances in convolution neural network (CNN) have fostered the progress in object recognition and semantic segmentation, which in turn has improved the performance of hyperspectral image (HSI) classification. Nevertheless, the difficulty of high dimensional feature extraction and the shortage of small training samples seriously hinder the future development of HSI classification. In this paper, we propose a novel algorithm for HSI classification based on three-dimensional (3D) CNN and a feature pyramid network (FPN), called 3D-FPN. The framework contains a principle component analysis, a feature extraction structure and a logistic regression. Specifically, the FPN built with 3D convolutions not only retains the advantages of 3D convolution to fully extract the spectral-spatial feature maps, but also concentrates on more detailed information and performs multi-scale feature fusion. This method avoids the excessive complexity of the model and is suitable for small sample hyperspectral classification with varying categories and spatial resolutions. In order to test the performance of our proposed 3D-FPN method, rigorous experimental analysis was performed on three public hyperspectral data sets and hyperspectral data of GF-5 satellite. Quantitative and qualitative results indicated that our proposed method attained the best performance among other current state-of-the-art end-to-end deep learning-based methods.
