Named Entity Recognition(NER)stands as a fundamental task within the field of biomedical text mining,aiming to extract specific types of entities such as genes,proteins,and diseases from complex biomedical texts and c...Named Entity Recognition(NER)stands as a fundamental task within the field of biomedical text mining,aiming to extract specific types of entities such as genes,proteins,and diseases from complex biomedical texts and categorize them into predefined entity types.This process can provide basic support for the automatic construction of knowledge bases.In contrast to general texts,biomedical texts frequently contain numerous nested entities and local dependencies among these entities,presenting significant challenges to prevailing NER models.To address these issues,we propose a novel Chinese nested biomedical NER model based on RoBERTa and Global Pointer(RoBGP).Our model initially utilizes the RoBERTa-wwm-ext-large pretrained language model to dynamically generate word-level initial vectors.It then incorporates a Bidirectional Long Short-Term Memory network for capturing bidirectional semantic information,effectively addressing the issue of long-distance dependencies.Furthermore,the Global Pointer model is employed to comprehensively recognize all nested entities in the text.We conduct extensive experiments on the Chinese medical dataset CMeEE and the results demonstrate the superior performance of RoBGP over several baseline models.This research confirms the effectiveness of RoBGP in Chinese biomedical NER,providing reliable technical support for biomedical information extraction and knowledge base construction.展开更多
Aluminosilicate small pore zeolites belonging to ABC-6 family play crucially important roles in the high methanol conversion with the high selectivity of light olefins,gas separation and storage,and selective catalyti...Aluminosilicate small pore zeolites belonging to ABC-6 family play crucially important roles in the high methanol conversion with the high selectivity of light olefins,gas separation and storage,and selective catalytic reduction of NO_(x).In this work,we report a general method,called the epitaxial growth approach,for designing ABC-6 family small pore zeolites.It is mainly realized through the epitaxial growth on the nonporous SOD-type zeolite in the presence of inorganic cations(Na^(+)and K^(+))combined with a variety of organic structure directing agents(OSDAs).In this case,a series of ABC-6 family small pore zeolites such as ERI-,SWY-,LEV-,AFX-,and PTT-type zeolites have been successfully synthesized within a few hours.More importantly,the advanced focused ion beam(FIB)and the low-dose high-resolution transmission electron microscopy(HRTEM)imaging technique have been utilized for unraveling the zeolite heterojunction at the atomic level during the epitaxial growth process.It turns out(222)crystallographic planes of the SOD-type zeolite substrate provide unique pre-building units,which facilitate the growth of targeted ABC-6 family small pore zeolites along its c-axis.Moreover,the morphologies of ERI-type zeolite can also be tuned through the epitaxial growth approach,achieving a longer lifetime in the methanol conversion.展开更多
In recent years,sensor technology has been widely used in the defense and control of sensitive areas in cities,or in various scenarios such as early warning of forest fires,monitoring of forest pests and diseases,and ...In recent years,sensor technology has been widely used in the defense and control of sensitive areas in cities,or in various scenarios such as early warning of forest fires,monitoring of forest pests and diseases,and protection of endangered animals.Deploying sensors to collect data and then utilizing unmanned aerial vehicle(UAV)to collect the data stored in the sensors has replaced traditional manual data collection as the dominant method.The current strategies for efficient data collection in above scenarios are still imperfect,and the low quality of the collected data and the excessive energy consumed by UAV flights are still the main problems faced in data collection.With regards this,this paper proposes a multi-UAV mission planning method for self-organized sensor data acquisition by comprehensively utilizing the techniques of self-organized sensor clustering,multi-UAV mission area allocation,and sub-area data acquisition scheme optimization.The improvedα-hop clustering method utilizes the average transmission distance to reduce the size of the collection sensors,and the K-Dimensional method is used to form a multi-UAV cooperative workspace,and then,the genetic algorithm is used to trade-off the speed with the age of information(AoI)of the collected information and the energy consumption to form the multi-UAV data collection operation scheme.The combined optimization scheme in paper improves the performance by 95.56%and 58.21%,respectively,compared to the traditional baseline model.In order to verify the excellent generalization and applicability of the proposed method in real scenarios,the simulation test is conducted by introducing the digital elevation model data of the real terrain,and the results show that the relative error values of the proposed method and the performance test of the actual flight of the UAV are within the error interval of±10%.Then,the advantages and disadvantages of the present method with the existing mainstream schemes are tested,and the results show that the present method has a huge advantage in terms of space and time complexity,and at the same time,the accuracy for data extraction is relatively improved by 10.46%and 12.71%.Finally,by eliminating the clustering process and the subtask assignment process,the AoI performance decreases by 3.46×and 4.45×,and the energy performance decreases by 3.52×and 4.47×.This paper presents a comprehensive and detailed proactive optimization of the existing challenges faced in the field of data acquisition by means of a series of combinatorial optimizations.展开更多
in the field of food flavor,there exists a substantial amount of structured and unstructured data originating from flavoromics,databases,and social media.to effectively extract valuable information from these diverse ...in the field of food flavor,there exists a substantial amount of structured and unstructured data originating from flavoromics,databases,and social media.to effectively extract valuable information from these diverse data sources and promote rational application,extensive data mining efforts have been undertaken.this review provides a systematic overview of data mining in the context of food flavor and summarizes various multivariate data processing strategies.this review examines a wide array of current research in flavoromics and discusses pre-processing methods designed to address challenges such as small dataset sizes and complex manual data preparation.furthermore,this review summarizes innovative approaches based on artificial intelligence and large language models,elucidating their prospective applications in flavor molecule prediction and recipe development.Lastly,we discuss the challenges and opportunities of applying data mining to flavor research.展开更多
Deep learning,a core branch of artificial intelligence,has shown great potential in food flavor analysis,prediction and optimization with its powerful data processing and pattern recognition capabilities.this article ...Deep learning,a core branch of artificial intelligence,has shown great potential in food flavor analysis,prediction and optimization with its powerful data processing and pattern recognition capabilities.this article reviews deep learning applications in food flavor,discussing various deep learning algorithms and models including artificial neural network,convolutional neural network,recurrent neural network,AutoEncoder,graph neural network,and generative adversarial network.besides,the latest progress and development trends of deep learning are discussed in this field.Compared with traditional flavor analysis methods,deep learning methods have obvious advantages and important application prospects in the field of food flavor.With the continuous advancement of technology in the future,it is expected that more deep learning applications will appear in the food industry.展开更多
Long-term traffic flow prediction is a crucial component of intelligent transportation systems within intelligent networks,requiring predictive models that balance accuracy with low-latency and lightweight computation...Long-term traffic flow prediction is a crucial component of intelligent transportation systems within intelligent networks,requiring predictive models that balance accuracy with low-latency and lightweight computation to optimize trafficmanagement and enhance urban mobility and sustainability.However,traditional predictivemodels struggle to capture long-term temporal dependencies and are computationally intensive,limiting their practicality in real-time.Moreover,many approaches overlook the periodic characteristics inherent in traffic data,further impacting performance.To address these challenges,we introduce ST-MambaGCN,a State-Space-Based Spatio-Temporal Graph Convolution Network.Unlike conventionalmodels,ST-MambaGCN replaces the temporal attention layer withMamba,a state-space model that efficiently captures long-term dependencies with near-linear computational complexity.The model combines Chebyshev polynomial-based graph convolutional networks(GCN)to explore spatial correlations.Additionally,we incorporate a multi-temporal feature capture mechanism,where the final integrated features are generated through the Hadamard product based on learnable parameters.This mechanism explicitly models shortterm,daily,and weekly traffic patterns to enhance the network’s awareness of traffic periodicity.Extensive experiments on the PeMS04 and PeMS08 datasets demonstrate that ST-MambaGCN significantly outperforms existing benchmarks,offering substantial improvements in both prediction accuracy and computational efficiency for long-term traffic flow prediction.展开更多
The Qinghai-Tibetan Plateau experienced a unique geological evolution during the Jurassic,driven by the termination of the Palaeotethys and the reduction of the Neotethys.The Indian Plate separated from the northern m...The Qinghai-Tibetan Plateau experienced a unique geological evolution during the Jurassic,driven by the termination of the Palaeotethys and the reduction of the Neotethys.The Indian Plate separated from the northern margin of Gondwana and drifted northward from the Southern Hemisphere.Given that the timing of strata serves as the basis for reconstructing geological history,the present work aimed to develop a new multiple stratigraphic and chronologic framework for the Jurassic strata of the Qinghai-Tibetan Plateau region via a synthesis of the material on lithostratigraphy,palaeontology,iso-radiometric dating,magnetostratigraphy,and other techniques with an emphasis on recent progress and findings.The new framework included the Jurassic System from the four major subdivisions of the plateau:the Baryan Har,Qiangtang,Lhasa-Gandise,and Southern Xizang(Himalaya).Ultimately,a more complete,refined biostratigraphic sequence was proposed,comprising the most common fossils in the plateau and those that are stratigraphically significant for the Jurassic stratigraphy,including ammonites,bivalves,brachiopods,foraminifera,radiolarians,and dinoflagellate cysts for the marine strata,and pollen and spores,and charophytes for the terrestrial sediments.This biostratigraphic framework was correlated with the Jurassic international standard zonation of the Geological Time Scale 2020 via standard or representative species or genera of ammonites.Based on this framework,we constructed a lateral correlation of the Jurassic strata between different basins of the plateau.The palaeontologic correlation in the present work shows that the Lhasa-Gandise Block had a closer relationship with the Qiangtang Block than with the Southern Xizang Himalaya during the Jurassic Period.Meanwhile,the Lhasa-Gandise Block and Qiangtang Block shared similar marine fauna features of the north marginal East Tethys.This contrasts the opinion suggesting that the Yarlung Zangbo Tethys was a small back-arc basin.A combination of stratigraphical,palaeontological,and sedimentological analyses implies that the Bangong Co-Nujiang Tethys may have begun rifting in the Late Triassic,evolving to the birth at the late Early Jurassic with the formation of ocean crust.However,this resulted in failure after it grew into the climax at the end of the Middle Jurassic when the Qiangtang Block began subducting under the Lhasa-Gandise Block.In the Early Cretaceous,the two blocks finally merged.展开更多
基金supported by the Outstanding Youth Team Project of Central Universities(QNTD202308)the Ant Group through CCF-Ant Research Fund(CCF-AFSG 769498 RF20220214).
文摘Named Entity Recognition(NER)stands as a fundamental task within the field of biomedical text mining,aiming to extract specific types of entities such as genes,proteins,and diseases from complex biomedical texts and categorize them into predefined entity types.This process can provide basic support for the automatic construction of knowledge bases.In contrast to general texts,biomedical texts frequently contain numerous nested entities and local dependencies among these entities,presenting significant challenges to prevailing NER models.To address these issues,we propose a novel Chinese nested biomedical NER model based on RoBERTa and Global Pointer(RoBGP).Our model initially utilizes the RoBERTa-wwm-ext-large pretrained language model to dynamically generate word-level initial vectors.It then incorporates a Bidirectional Long Short-Term Memory network for capturing bidirectional semantic information,effectively addressing the issue of long-distance dependencies.Furthermore,the Global Pointer model is employed to comprehensively recognize all nested entities in the text.We conduct extensive experiments on the Chinese medical dataset CMeEE and the results demonstrate the superior performance of RoBGP over several baseline models.This research confirms the effectiveness of RoBGP in Chinese biomedical NER,providing reliable technical support for biomedical information extraction and knowledge base construction.
基金supported by National Key Research and Development Project of China(No.2022YFE0113800)National Natural Science Foundation of China(Nos.22288101,21972136,21991090 and 21991091)Key Research Program of Frontier Sciences,Chinese Academy of Sciences(No.QYZDB-SSW-JSC040)。
文摘Aluminosilicate small pore zeolites belonging to ABC-6 family play crucially important roles in the high methanol conversion with the high selectivity of light olefins,gas separation and storage,and selective catalytic reduction of NO_(x).In this work,we report a general method,called the epitaxial growth approach,for designing ABC-6 family small pore zeolites.It is mainly realized through the epitaxial growth on the nonporous SOD-type zeolite in the presence of inorganic cations(Na^(+)and K^(+))combined with a variety of organic structure directing agents(OSDAs).In this case,a series of ABC-6 family small pore zeolites such as ERI-,SWY-,LEV-,AFX-,and PTT-type zeolites have been successfully synthesized within a few hours.More importantly,the advanced focused ion beam(FIB)and the low-dose high-resolution transmission electron microscopy(HRTEM)imaging technique have been utilized for unraveling the zeolite heterojunction at the atomic level during the epitaxial growth process.It turns out(222)crystallographic planes of the SOD-type zeolite substrate provide unique pre-building units,which facilitate the growth of targeted ABC-6 family small pore zeolites along its c-axis.Moreover,the morphologies of ERI-type zeolite can also be tuned through the epitaxial growth approach,achieving a longer lifetime in the methanol conversion.
基金National Key R&D Program of China(2022YFF1302700)Xiong’an New Area Science and Technology Innovation Special Project of Ministry of Science and Technology of China(2023XAGG0065)+2 种基金Ant Group through CCF-Ant Research Fund(CCF-AFSG RF20220214)Outstanding Youth Team Project of Central Universities(QNTD202308)Beijing Forestry University National Training Program of Innovation and Entrepreneurship for Undergraduates(202310022097).
文摘In recent years,sensor technology has been widely used in the defense and control of sensitive areas in cities,or in various scenarios such as early warning of forest fires,monitoring of forest pests and diseases,and protection of endangered animals.Deploying sensors to collect data and then utilizing unmanned aerial vehicle(UAV)to collect the data stored in the sensors has replaced traditional manual data collection as the dominant method.The current strategies for efficient data collection in above scenarios are still imperfect,and the low quality of the collected data and the excessive energy consumed by UAV flights are still the main problems faced in data collection.With regards this,this paper proposes a multi-UAV mission planning method for self-organized sensor data acquisition by comprehensively utilizing the techniques of self-organized sensor clustering,multi-UAV mission area allocation,and sub-area data acquisition scheme optimization.The improvedα-hop clustering method utilizes the average transmission distance to reduce the size of the collection sensors,and the K-Dimensional method is used to form a multi-UAV cooperative workspace,and then,the genetic algorithm is used to trade-off the speed with the age of information(AoI)of the collected information and the energy consumption to form the multi-UAV data collection operation scheme.The combined optimization scheme in paper improves the performance by 95.56%and 58.21%,respectively,compared to the traditional baseline model.In order to verify the excellent generalization and applicability of the proposed method in real scenarios,the simulation test is conducted by introducing the digital elevation model data of the real terrain,and the results show that the relative error values of the proposed method and the performance test of the actual flight of the UAV are within the error interval of±10%.Then,the advantages and disadvantages of the present method with the existing mainstream schemes are tested,and the results show that the present method has a huge advantage in terms of space and time complexity,and at the same time,the accuracy for data extraction is relatively improved by 10.46%and 12.71%.Finally,by eliminating the clustering process and the subtask assignment process,the AoI performance decreases by 3.46×and 4.45×,and the energy performance decreases by 3.52×and 4.47×.This paper presents a comprehensive and detailed proactive optimization of the existing challenges faced in the field of data acquisition by means of a series of combinatorial optimizations.
基金supported by the Jiangsu Specially-Appointed Professor Program(5966010241250010)the fundamental research funds for the Central Universities(JUSRP124032).
文摘in the field of food flavor,there exists a substantial amount of structured and unstructured data originating from flavoromics,databases,and social media.to effectively extract valuable information from these diverse data sources and promote rational application,extensive data mining efforts have been undertaken.this review provides a systematic overview of data mining in the context of food flavor and summarizes various multivariate data processing strategies.this review examines a wide array of current research in flavoromics and discusses pre-processing methods designed to address challenges such as small dataset sizes and complex manual data preparation.furthermore,this review summarizes innovative approaches based on artificial intelligence and large language models,elucidating their prospective applications in flavor molecule prediction and recipe development.Lastly,we discuss the challenges and opportunities of applying data mining to flavor research.
基金supported by the Jiangsu Specially-Appointed Professor Program(5966010241250010)the fundamental research funds for the Central Universities(JUSRP124032).
文摘Deep learning,a core branch of artificial intelligence,has shown great potential in food flavor analysis,prediction and optimization with its powerful data processing and pattern recognition capabilities.this article reviews deep learning applications in food flavor,discussing various deep learning algorithms and models including artificial neural network,convolutional neural network,recurrent neural network,AutoEncoder,graph neural network,and generative adversarial network.besides,the latest progress and development trends of deep learning are discussed in this field.Compared with traditional flavor analysis methods,deep learning methods have obvious advantages and important application prospects in the field of food flavor.With the continuous advancement of technology in the future,it is expected that more deep learning applications will appear in the food industry.
基金supported byNationalNatural Science Foundation of China,GrantNo.62402046the Beijing Forestry University Science and Technology Innovation Project under Grant No.BLX202358.
文摘Long-term traffic flow prediction is a crucial component of intelligent transportation systems within intelligent networks,requiring predictive models that balance accuracy with low-latency and lightweight computation to optimize trafficmanagement and enhance urban mobility and sustainability.However,traditional predictivemodels struggle to capture long-term temporal dependencies and are computationally intensive,limiting their practicality in real-time.Moreover,many approaches overlook the periodic characteristics inherent in traffic data,further impacting performance.To address these challenges,we introduce ST-MambaGCN,a State-Space-Based Spatio-Temporal Graph Convolution Network.Unlike conventionalmodels,ST-MambaGCN replaces the temporal attention layer withMamba,a state-space model that efficiently captures long-term dependencies with near-linear computational complexity.The model combines Chebyshev polynomial-based graph convolutional networks(GCN)to explore spatial correlations.Additionally,we incorporate a multi-temporal feature capture mechanism,where the final integrated features are generated through the Hadamard product based on learnable parameters.This mechanism explicitly models shortterm,daily,and weekly traffic patterns to enhance the network’s awareness of traffic periodicity.Extensive experiments on the PeMS04 and PeMS08 datasets demonstrate that ST-MambaGCN significantly outperforms existing benchmarks,offering substantial improvements in both prediction accuracy and computational efficiency for long-term traffic flow prediction.
基金supported by the Second Tibetan Plateau Scientific Expedition and Research(Grant No.2019QZKK0706)the National Natural Science Foundation of China(Grant Nos.42372019,41888101,41872004,42272027,42288201,42172028)the Strategic Priority Research Program of the Chinese Academy of Sciences(Grant Nos.XDB26000000,XDA2007020203)。
文摘The Qinghai-Tibetan Plateau experienced a unique geological evolution during the Jurassic,driven by the termination of the Palaeotethys and the reduction of the Neotethys.The Indian Plate separated from the northern margin of Gondwana and drifted northward from the Southern Hemisphere.Given that the timing of strata serves as the basis for reconstructing geological history,the present work aimed to develop a new multiple stratigraphic and chronologic framework for the Jurassic strata of the Qinghai-Tibetan Plateau region via a synthesis of the material on lithostratigraphy,palaeontology,iso-radiometric dating,magnetostratigraphy,and other techniques with an emphasis on recent progress and findings.The new framework included the Jurassic System from the four major subdivisions of the plateau:the Baryan Har,Qiangtang,Lhasa-Gandise,and Southern Xizang(Himalaya).Ultimately,a more complete,refined biostratigraphic sequence was proposed,comprising the most common fossils in the plateau and those that are stratigraphically significant for the Jurassic stratigraphy,including ammonites,bivalves,brachiopods,foraminifera,radiolarians,and dinoflagellate cysts for the marine strata,and pollen and spores,and charophytes for the terrestrial sediments.This biostratigraphic framework was correlated with the Jurassic international standard zonation of the Geological Time Scale 2020 via standard or representative species or genera of ammonites.Based on this framework,we constructed a lateral correlation of the Jurassic strata between different basins of the plateau.The palaeontologic correlation in the present work shows that the Lhasa-Gandise Block had a closer relationship with the Qiangtang Block than with the Southern Xizang Himalaya during the Jurassic Period.Meanwhile,the Lhasa-Gandise Block and Qiangtang Block shared similar marine fauna features of the north marginal East Tethys.This contrasts the opinion suggesting that the Yarlung Zangbo Tethys was a small back-arc basin.A combination of stratigraphical,palaeontological,and sedimentological analyses implies that the Bangong Co-Nujiang Tethys may have begun rifting in the Late Triassic,evolving to the birth at the late Early Jurassic with the formation of ocean crust.However,this resulted in failure after it grew into the climax at the end of the Middle Jurassic when the Qiangtang Block began subducting under the Lhasa-Gandise Block.In the Early Cretaceous,the two blocks finally merged.
