We proposes an AI-assisted framework for integrated natural disaster prevention and emergency response,leveraging the DeepSeek large language model(LLM)to advance intelligent decision-making in geohazard management.We...We proposes an AI-assisted framework for integrated natural disaster prevention and emergency response,leveraging the DeepSeek large language model(LLM)to advance intelligent decision-making in geohazard management.We systematically analyze the technical pathways for deploying LLMs in disaster scenarios,emphasizing three breakthrough directions:(1)knowledge graph-driven dynamic risk modeling,(2)reinforcement learning-optimized emergency decision systems,and(3)secure local deployment architectures.The DeepSeek model demonstrates unique advantages through its hybrid reasoning mechanism combining semantic analysis with geospatial pattern recognition,enabling cost-effective processing of multi-source data spanning historical disaster records,real-time IoT sensor feeds,and socio-environmental parameters.A modular system architecture is designed to achieve three critical objectives:(a)automated construction of domain-specific knowledge graphs through unsupervised learning of disaster physics relationships,(b)scenario-adaptive resource allocation using risk simulations,and(c)preserving emergency coordination via federated learning across distributed response nodes.The proposed local deployment paradigm addresses critical data security concerns in cross-border disaster management while complying with the FAIR principles(Findable,Accessible,Interoperable,Reusable)for geoscientific data governance.This work establishes a methodological foundation for next-generation AI-earth science convergence in disaster mitigation.展开更多
The aim of this article is to explore potential directions for the development of artificial intelligence(AI).It points out that,while current AI can handle the statistical properties of complex systems,it has difficu...The aim of this article is to explore potential directions for the development of artificial intelligence(AI).It points out that,while current AI can handle the statistical properties of complex systems,it has difficulty effectively processing and fully representing their spatiotemporal complexity patterns.The article also discusses a potential path of AI development in the engineering domain.Based on the existing understanding of the principles of multilevel com-plexity,this article suggests that consistency among the logical structures of datasets,AI models,model-building software,and hardware will be an important AI development direction and is worthy of careful consideration.展开更多
DeepSeek,a Chinese artificial intelligence(AI)startup,has released their V3 and R1 series models,which attracted global attention due to their low cost,high performance,and open-source advantages.This paper begins by ...DeepSeek,a Chinese artificial intelligence(AI)startup,has released their V3 and R1 series models,which attracted global attention due to their low cost,high performance,and open-source advantages.This paper begins by reviewing the evolution of large AI models focusing on paradigm shifts,the mainstream large language model(LLM)paradigm,and the DeepSeek paradigm.Subsequently,the paper highlights novel algorithms introduced by DeepSeek,including multi-head latent attention(MLA),mixture-of-experts(MoE),multi-token prediction(MTP),and group relative policy optimization(GRPO).The paper then explores DeepSeek's engineering breakthroughs in LLM scaling,training,inference,and system-level optimization architecture.Moreover,the impact of DeepSeek models on the competitive AI landscape is analyzed,comparing them to mainstream LLMs across various fields.Finally,the paper reflects on the insights gained from DeepSeek's innovations and discusses future trends in the technical and engineering development of large AI models,particularly in data,training,and reasoning.展开更多
This study explores the determinants of impact on ecology in Northern Tanzania.By examining key socio-economic,institutional,and structural factors influencing engagement the study provides insights in strengthening a...This study explores the determinants of impact on ecology in Northern Tanzania.By examining key socio-economic,institutional,and structural factors influencing engagement the study provides insights in strengthening agribusiness networks and improving livelihoods.Data was collected from 215 farmers and 320 traders through a multistage sampling procedure.Heckman AI sample selection model was used in data analysis whereby the findings showed key factors influencing farmers’decisions on ecology were gender and years of formal education at p<0.1,and access to finance and off-farm income at p<0.05.The degree of farmers participation in social groups was influenced by age,household size,off-farm income and business network at p<0.05,number of years in formal education and access to finance at p<0.01,and distance to the market at p<0.1.The decision of traders to impact on ecology was significantly influenced by age and trading experience at p<0.1.Meanwhile,the degree of their involvement in social groups was strongly affected by gender,formal education,and trust at p<0.01,as well as by access to finance and business networks at p<0.05.The study concluded that natural ecology is influenced by socio economic and structural factors but trust among group members determine the degree of participation.The study recommends that strategies to improve agribusiness networks should understand underlying causes of impact on ecology and strengthen available social groups to improve performance of farmers and traders.展开更多
With the rapid development of large AI models,large decision models have further broken through the limits of human cognition and promoted the innovation of decision-making paradigms in extensive fields such as medici...With the rapid development of large AI models,large decision models have further broken through the limits of human cognition and promoted the innovation of decision-making paradigms in extensive fields such as medicine and transportation.In this paper,we systematically expound on the intelligent decision-making technology and prospects driven by large AI models.Specifically,we first review the development of large AI models in recent years.Then,from the perspective of methods,we introduce important theories and technologies of large decision models,such as model architecture and model adaptation.Next,from the perspective of applications,we introduce the cutting-edge applications of large decision models in various fields,such as autonomous driving and knowledge decision-making.Finally,we discuss existing challenges,such as security issues,decision bias and hallucination phenomenon as well as future prospects,from both technology development and domain applications.We hope this review paper can help researchers understand the important progress of intelligent decision-making driven by large AI models.展开更多
Objective:The current pathological diagnosis of lymph node metastasis is time-consuming,labor-intensive,and dependent on sectioning of paraffin blocks.Herein,in a prospective cohort of patients with breast cancer,we v...Objective:The current pathological diagnosis of lymph node metastasis is time-consuming,labor-intensive,and dependent on sectioning of paraffin blocks.Herein,in a prospective cohort of patients with breast cancer,we validated dynamic full-field optical coherence tomography(D-FFOCT),a virtual pathology tool integrating deep learning for nodal metastasis detection,and offering rapid and label-free histologic approximations of fresh tissues.Methods:In a prospective dual-center cohort of 155 patients with breast cancer,747 freshly bisected lymph node slides were obtained via D-FFOCT.Surgeons interpreted each slide with histopathology as the gold standard.A deep learning model was trained on 28,911 patches(corresponding to 590 slides)and tested on 7,736 patches(corresponding to 157 slides).The results were mapped to the slide level for potential intraoperative evaluation.Results:D-FFOCT strongly correlated with hematoxylin and eosin(H&E)-stained histological images.Surgeons achieved 97.10%specificity in nodal diagnosis with D-FFOCT.The performance of the artificial intelligence(AI)model was not inferior to that of human experts and had a sensitivity/specificity of 87.88%/91.94%and an area under the receiver operating characteristic curve of 0.899 at the slide level.The human–AI collaborative system reduced labor requirements by 75%and increased the specificity by 6.5%,to 98.39%.Conclusions:D-FFOCT has excellent potential as a tool for assessing lymph node metastatic status without tissue preparation or consumption.The integration of D-FFOCT with deep learning decreases labor demands and maintains high accuracy,thereby enabling streamlined nodal prediction independent of routine pathology procedures.展开更多
Artificial intelligence(AI)models are promising to improve the accuracy of wireless positioning systems,particularly in indoor environments where unpredictable radio propagation channel is a great challenge.Although g...Artificial intelligence(AI)models are promising to improve the accuracy of wireless positioning systems,particularly in indoor environments where unpredictable radio propagation channel is a great challenge.Although great efforts have been made to explore the effectiveness of different AI models,it is still an open problem whether these models,trained with the data collected from all base stations(BSs),could work when some BSs are unavailable.In this paper,we make the first effort to enhance the generalization ability of AI wireless positioning model to adapt to the scenario where only partial BSs work.Particularly,a Siamese Network based Wireless Positioning Model(SNWPM)is proposed to predict the location of mobile user equipment from channel state information(CSI)collected from 5G BSs.Furthermore,a Feature Aware Attention Module(FAAM)is introduced to reinforce the capability of feature extraction from CSI data.Experiments are conducted on the 2022 Wireless Communication AI Competition(WAIC)dataset.The proposed SNWPM achieves decimeter-level positioning accuracy even if the data of partial BSs are unavailable.Compared with other AI models,the proposed SNWPM can reduce the positioning error by nearly 50%to more than 60%while using less parameters and lower computation resources.展开更多
In a prior practice and policy article published in Healthcare Science,we introduced the deployed application of an artificial intelligence(AI)model to predict longer‐term inpatient readmissions to guide community ca...In a prior practice and policy article published in Healthcare Science,we introduced the deployed application of an artificial intelligence(AI)model to predict longer‐term inpatient readmissions to guide community care interventions for patients with complex conditions in the context of Singapore's Hospital to Home(H2H)program that has been operating since 2017.In this follow on practice and policy article,we further elaborate on Singapore's H2H program and care model,and its supporting AI model for multiple readmission prediction,in the following ways:(1)by providing updates on the AI and supporting information systems,(2)by reporting on customer engagement and related service delivery outcomes including staff‐related time savings and patient benefits in terms of bed days saved,(3)by sharing lessons learned with respect to(i)analytics challenges encountered due to the high degree of heterogeneity and resulting variability of the data set associated with the population of program participants,(ii)balancing competing needs for simpler and stable predictive models versus continuing to further enhance models and add yet more predictive variables,and(iii)the complications of continuing to make model changes when the AI part of the system is highly interlinked with supporting clinical information systems,(4)by highlighting how this H2H effort supported broader Covid‐19 response efforts across Singapore's public healthcare system,and finally(5)by commenting on how the experiences and related capabilities acquired from running this H2H program and related community care model and supporting AI prediction model are expected to contribute to the next wave of Singapore's public healthcare efforts from 2023 onwards.For the convenience of the reader,some content that introduces the H2H program and the multiple readmissions AI prediction model that previously appeared in the prior Healthcare Science publication is repeated at the beginning of this article.展开更多
Robotic computing systems play an important role in enabling intelligent robotic tasks through intelligent algo-rithms and supporting hardware.In recent years,the evolution of robotic algorithms indicates a roadmap fr...Robotic computing systems play an important role in enabling intelligent robotic tasks through intelligent algo-rithms and supporting hardware.In recent years,the evolution of robotic algorithms indicates a roadmap from traditional robotics to hierarchical and end-to-end models.This algorithmic advancement poses a critical challenge in achieving balanced system-wide performance.Therefore,algorithm-hardware co-design has emerged as the primary methodology,which ana-lyzes algorithm behaviors on hardware to identify common computational properties.These properties can motivate algo-rithm optimization to reduce computational complexity and hardware innovation from architecture to circuit for high performance and high energy efficiency.We then reviewed recent works on robotic and embodied AI algorithms and computing hard-ware to demonstrate this algorithm-hardware co-design methodology.In the end,we discuss future research opportunities by answering two questions:(1)how to adapt the computing platforms to the rapid evolution of embodied AI algorithms,and(2)how to transform the potential of emerging hardware innovations into end-to-end inference improvements.展开更多
By comparing price plans offered by several retail energy firms,end users with smart meters and controllers may optimize their energy use cost portfolios,due to the growth of deregulated retail power markets.To help s...By comparing price plans offered by several retail energy firms,end users with smart meters and controllers may optimize their energy use cost portfolios,due to the growth of deregulated retail power markets.To help smart grid end-users decrease power payment and usage unhappiness,this article suggests a decision system based on reinforcement learning to aid with electricity price plan selection.An enhanced state-based Markov decision process(MDP)without transition probabilities simulates the decision issue.A Kernel approximate-integrated batch Q-learning approach is used to tackle the given issue.Several adjustments to the sampling and data representation are made to increase the computational and prediction performance.Using a continuous high-dimensional state space,the suggested approach can uncover the underlying characteristics of time-varying pricing schemes.Without knowing anything regarding the market environment in advance,the best decision-making policy may be learned via case studies that use data from actual historical price plans.Experiments show that the suggested decision approach may reduce cost and energy usage dissatisfaction by using user data to build an accurate prediction strategy.In this research,we look at how smart city energy planners rely on precise load forecasts.It presents a hybrid method that extracts associated characteristics to improve accuracy in residential power consumption forecasts using machine learning(ML).It is possible to measure the precision of forecasts with the use of loss functions with the RMSE.This research presents a methodology for estimating smart home energy usage in response to the growing interest in explainable artificial intelligence(XAI).Using Shapley Additive explanations(SHAP)approaches,this strategy makes it easy for consumers to comprehend their energy use trends.To predict future energy use,the study employs gradient boosting in conjunction with long short-term memory neural networks.展开更多
From globally popular video game Black Myth:Wukong,which has garnered a dedicated player base around the world,to DeepSeek,an artificial intelligence(AI)model developed at an impressively low cost that rivals U.S.comp...From globally popular video game Black Myth:Wukong,which has garnered a dedicated player base around the world,to DeepSeek,an artificial intelligence(AI)model developed at an impressively low cost that rivals U.S.company OpenAI’s ChatGPT,and the perfectly synchronized robotic ensemble performing with precision at this year’s China Central Television Spring Festival Gala,a Chinese New Year’s Eve extravaganza that aired on January 28-these big tech breakthroughs have risen to prominence one after another,generating massive buzz.展开更多
In recent years,the rapid advancement of artificial intelligence(AI)has fostered deep integration between large AI models and robotic technology.Robots such as robotic dogs capable of carrying heavy loads on mountaino...In recent years,the rapid advancement of artificial intelligence(AI)has fostered deep integration between large AI models and robotic technology.Robots such as robotic dogs capable of carrying heavy loads on mountainous terrain or performing waste disposal tasks and humanoid robots that can execute high-precision component installations have gradually reached the public eye,raising expectations for embodied intelligent robots.展开更多
Time series anomaly detection is critical in domains such as manufacturing,finance,and cybersecurity.Recent generative AI models,particularly Transformer-and Autoencoder-based architectures,show strong accuracy but th...Time series anomaly detection is critical in domains such as manufacturing,finance,and cybersecurity.Recent generative AI models,particularly Transformer-and Autoencoder-based architectures,show strong accuracy but their robustness under noisy conditions is less understood.This study evaluates three representative models—AnomalyTransformer,TranAD,and USAD—on the Server Machine Dataset(SMD)and cross-domain benchmarks including the SoilMoisture Active Passive(SMAP)dataset,theMars Science Laboratory(MSL)dataset,and the Secure Water Treatment(SWaT)testbed.Seven noise settings(five canonical,two mixed)at multiple intensities are tested under fixed clean-data training,with variations in window,stride,and thresholding.Results reveal distinct robustness profiles:AnomalyTransformermaintains recall but loses precision under abrupt noise,TranAD balances sensitivity yet is vulnerable to structured anomalies,and USAD resists Gaussian perturbations but collapses under block anomalies.Quantitatively,F1 drops 60%–70%on noisy SMD,with severe collapse in SWaT(F1≤0.10,Drop up to 84%)but relative stability on SMAP/MSL(Drop within±10%).Overall,generative models exhibit complementary robustness patterns,highlighting noise-type dependent vulnerabilities and providing practical guidance for robust deployment.展开更多
With the exponential growth in the complexity of Internet architectures and the widespread adoption of cloud-native service technologies,traditional operation and maintenance(Artificial Intelligence for IT Operations,...With the exponential growth in the complexity of Internet architectures and the widespread adoption of cloud-native service technologies,traditional operation and maintenance(Artificial Intelligence for IT Operations,AIOps)models-largely reliant on the paradigm of"expert rules+fixed scripts"-have become increasingly passive and inefficient when confronted with unknown faults and massive volumes of alerts.This study focuses on the application of large-scale AI model-based intelligent agents across the full lifecycle of Internet fault handling,aiming to construct autonomous O&M agents endowed with capabilities of perception,decision-making,and execution.The paper first analyzes the core challenges in current fault management:alert storms leading to missed and false incident reports,cross-system data silos hindering root cause localization,and heavy reliance on expert experience in manual troubleshooting,resulting in delayed response times.On this basis,a hierarchical solution architecture based on large-model agents is proposed,comprising a multi-source data perception layer,a fault reasoning and decision-making layer,and an automated execution layer[1].By integrating Retrieval-Augmented Generation(RAG)techniques with an O&M knowledge base,the proposed approach equips intelligent agents with the ability to interpret topology metrics,log semantics,and change events.Furthermore,the introduction of chain-of-thought reasoning and reflection mechanisms enables the agents to simulate expert diagnostic pathways,thereby achieving millisecond-level anomaly detection and minute-level root cause identification.展开更多
Soil permeability is a critical parameter that dictates the movement of water through soil,and it impacts processes such as seepage,erosion,slope stability,foundation design,groundwater contamination,and various engin...Soil permeability is a critical parameter that dictates the movement of water through soil,and it impacts processes such as seepage,erosion,slope stability,foundation design,groundwater contamination,and various engineering applications.This study investigates the permeability of soil amended with waste foundry sand(WFS)at a replacement level of 10%.Permeability measurements are conducted for three distinct relative densities,spanning from 65% to 85%.The dataset compiled from these measurements is employed to develop ensemble artificial intelligence(AI)models.Specifically,four regressor AI models are considered:Nearest Neighbor(NNR),Decision Tree(DTR),Random Forest(RFR)and Support Vector Machine(SVR).These models are enhanced with four distinct base learners:Gradient Boosting(GB),Stacking Regressor(SR),AdaBoost Regressor(ADR),and XGBoost(XGB).The input parameters include fraction of base sand(BS),fraction of waste foundry sand(WFS),relative density(RD),duration of flow(T),quantity of flow(Q)and permeability(k),totalling 165 data points.Through comparative analysis,the Gradient Boost with Decision Tree(GB-DTR)model is found to be best-performed model,with R2=0.9919.Sensitivity analysis reveals that Q is the most influential input parameter in predicting soil permeability.展开更多
With the rapid progress of AI technology,AI big models with transformer architecture as the core has made great progress in natural language processing,computer vision and other fields.Smart grid is a modern power sys...With the rapid progress of AI technology,AI big models with transformer architecture as the core has made great progress in natural language processing,computer vision and other fields.Smart grid is a modern power system integrated with advanced information,communication and control technology.The complexity and variability of the system and the massive reference data provide application scenarios for the application of AI large model.This paper systematically expounds the key technologies matching with AI large model and its adaptability to the core links of smart grid,and focuses on the role of AI large model in smart grid construction,such as new energy grid connection,equipment management,grid topology optimization and dispatching decision,such as specific application modes and cases in load forecasting,real-time dispatching and multi-energy complementarity.At the same time,this paper deeply analyzes the key challenges in data,technology,engineering and security faced by the application of AI large model in various fields of power,and puts forward the corresponding optimal solutions.Finally,combined with typical cases,the future development direction of the integration of digital twins,generative AI and other technologies is conceived,which provides a theoretical reference and practical path for promoting the autonomous and efficient development of smart grid.展开更多
Rechargeable batteries are pivotal for achieving carbon neutrality and enabling the renewable energy transition.Their advancement requires inno-vations at micro(materials),device(manufacturing),and system(control and ...Rechargeable batteries are pivotal for achieving carbon neutrality and enabling the renewable energy transition.Their advancement requires inno-vations at micro(materials),device(manufacturing),and system(control and optimization)levels.However,traditional trial-and-error approaches are inadequate for modern scientific demands.As a transformative artificial intelligence(AI)technology,large language models(LLMs)deliver powerful semantic understanding and reasoning capabilities,driving a paradigm shift in battery research to address multilevel innovation needs.Neverthe-less,this field still faces dual challenges:ambiguous technical roadmaps and fragmented progress in stage-specific achievements.This review sys-tematically consolidates recent advances in applying LLMs to battery research,distilling core findings across four critical domains:knowledge integration,materials discovery,manufacturing processes,and system management.To address key bottlenecks—including limited model inter-pretability,inadequate alignment with electrochemical mechanisms,and real-world data adaptation challenges—we propose structured frameworks for deep integration of battery research and LLMs,alongside defined future technical pathways.These frameworks bridge fundamental battery science with AI-driven innovation paradigms to facilitate groundbreaking advances in next-generation battery technologies.展开更多
As modern communication technology advances apace,the digital communication signals identification plays an important role in cognitive radio networks,the communication monitoring and management systems.AI has become ...As modern communication technology advances apace,the digital communication signals identification plays an important role in cognitive radio networks,the communication monitoring and management systems.AI has become a promising solution to this problem due to its powerful modeling capability,which has become a consensus in academia and industry.However,because of the data-dependence and inexplicability of AI models and the openness of electromagnetic space,the physical layer digital communication signals identification model is threatened by adversarial attacks.Adversarial examples pose a common threat to AI models,where well-designed and slight perturbations added to input data can cause wrong results.Therefore,the security of AI models for the digital communication signals identification is the premise of its efficient and credible applications.In this paper,we first launch adversarial attacks on the end-to-end AI model for automatic modulation classifi-cation,and then we explain and present three defense mechanisms based on the adversarial principle.Next we present more detailed adversarial indicators to evaluate attack and defense behavior.Finally,a demonstration verification system is developed to show that the adversarial attack is a real threat to the digital communication signals identification model,which should be paid more attention in future research.展开更多
Generating carbon credits in rural and wetland lagoon environments is important for the economic and social survival of the same.There are many methodologies to study and certificate the Carbon Sink such as the ISO 14...Generating carbon credits in rural and wetland lagoon environments is important for the economic and social survival of the same.There are many methodologies to study and certificate the Carbon Sink such as the ISO 14064,VCS VERRA,UNI-BNEUTRAL,GOLD STANDARD and others.Many methods done before 2018 are obsolete since research has developed greatly in recent years.The methods are all different,but they share a continuous and real monitoring of the environment to ensure a true CCS(Carbon Capture and Storage)action.In the case of absence of monitoring,the method uses a system of provision of carbon credits called“buffer”.This system allows maintaining a credit-generating activity even in the presence of important anomalies due to adverse weather events.This research shows the complex analytic web of the different sensors in a continuous environmental monitoring system via GSM(Global System for Mobile)Communication and IoT(Internet of Things).By 2011,a monitoring network was installed in the wetland environments of Northern Italy Venetian Lagoon(UNESCO heritage)and used to understand and validate,the CCS action.Thingspeak cloud platform is used to collect data and is used to send alert to the user if the biological sink is reversed to emission.The obtained large dataset was used to prepare a AI(Artificial Intelligence)model“CCS wetland forecast”by Google COLAB.This model can fit the trend to avoid the direct and spot chemical field analysis and demonstrate the real efficacy of the model chosen.This network is now implemented by the Italian national method UNI PdR 99:2021 BNeutral generation of carbon credits.展开更多
基金funded by the Chongqing Water Resources Bureau,China(Project No.CQS24C00836).
文摘We proposes an AI-assisted framework for integrated natural disaster prevention and emergency response,leveraging the DeepSeek large language model(LLM)to advance intelligent decision-making in geohazard management.We systematically analyze the technical pathways for deploying LLMs in disaster scenarios,emphasizing three breakthrough directions:(1)knowledge graph-driven dynamic risk modeling,(2)reinforcement learning-optimized emergency decision systems,and(3)secure local deployment architectures.The DeepSeek model demonstrates unique advantages through its hybrid reasoning mechanism combining semantic analysis with geospatial pattern recognition,enabling cost-effective processing of multi-source data spanning historical disaster records,real-time IoT sensor feeds,and socio-environmental parameters.A modular system architecture is designed to achieve three critical objectives:(a)automated construction of domain-specific knowledge graphs through unsupervised learning of disaster physics relationships,(b)scenario-adaptive resource allocation using risk simulations,and(c)preserving emergency coordination via federated learning across distributed response nodes.The proposed local deployment paradigm addresses critical data security concerns in cross-border disaster management while complying with the FAIR principles(Findable,Accessible,Interoperable,Reusable)for geoscientific data governance.This work establishes a methodological foundation for next-generation AI-earth science convergence in disaster mitigation.
文摘The aim of this article is to explore potential directions for the development of artificial intelligence(AI).It points out that,while current AI can handle the statistical properties of complex systems,it has difficulty effectively processing and fully representing their spatiotemporal complexity patterns.The article also discusses a potential path of AI development in the engineering domain.Based on the existing understanding of the principles of multilevel com-plexity,this article suggests that consistency among the logical structures of datasets,AI models,model-building software,and hardware will be an important AI development direction and is worthy of careful consideration.
基金supported by the National Natural Science Foundation of China(62233005,62293502,U2441245,62176185,U23B2057,62306112)the STCSM Science and Technology Innovation Action Plan Computational Biology Program(24JS2830400)+2 种基金the State Key Laboratory of Industrial Control Technology,China(ICT2024A22)the Shanghai Sailing Program(23YF1409400)the National Science and Technology Major Project(2024ZD0532403).
文摘DeepSeek,a Chinese artificial intelligence(AI)startup,has released their V3 and R1 series models,which attracted global attention due to their low cost,high performance,and open-source advantages.This paper begins by reviewing the evolution of large AI models focusing on paradigm shifts,the mainstream large language model(LLM)paradigm,and the DeepSeek paradigm.Subsequently,the paper highlights novel algorithms introduced by DeepSeek,including multi-head latent attention(MLA),mixture-of-experts(MoE),multi-token prediction(MTP),and group relative policy optimization(GRPO).The paper then explores DeepSeek's engineering breakthroughs in LLM scaling,training,inference,and system-level optimization architecture.Moreover,the impact of DeepSeek models on the competitive AI landscape is analyzed,comparing them to mainstream LLMs across various fields.Finally,the paper reflects on the insights gained from DeepSeek's innovations and discusses future trends in the technical and engineering development of large AI models,particularly in data,training,and reasoning.
基金financed as part of the project“Development of a methodology for instrumental base formation for analysis and modeling of the spatial socio-economic development of systems based on internal reserves in the context of digitalization”(FSEG-2023-0008).
文摘This study explores the determinants of impact on ecology in Northern Tanzania.By examining key socio-economic,institutional,and structural factors influencing engagement the study provides insights in strengthening agribusiness networks and improving livelihoods.Data was collected from 215 farmers and 320 traders through a multistage sampling procedure.Heckman AI sample selection model was used in data analysis whereby the findings showed key factors influencing farmers’decisions on ecology were gender and years of formal education at p<0.1,and access to finance and off-farm income at p<0.05.The degree of farmers participation in social groups was influenced by age,household size,off-farm income and business network at p<0.05,number of years in formal education and access to finance at p<0.01,and distance to the market at p<0.1.The decision of traders to impact on ecology was significantly influenced by age and trading experience at p<0.1.Meanwhile,the degree of their involvement in social groups was strongly affected by gender,formal education,and trust at p<0.01,as well as by access to finance and business networks at p<0.05.The study concluded that natural ecology is influenced by socio economic and structural factors but trust among group members determine the degree of participation.The study recommends that strategies to improve agribusiness networks should understand underlying causes of impact on ecology and strengthen available social groups to improve performance of farmers and traders.
基金supported by the National Natural Science Foundation of China(Grant 62293545)Shenzhen Science and Technology Program(Grant ZDSYS20220323112000001).
文摘With the rapid development of large AI models,large decision models have further broken through the limits of human cognition and promoted the innovation of decision-making paradigms in extensive fields such as medicine and transportation.In this paper,we systematically expound on the intelligent decision-making technology and prospects driven by large AI models.Specifically,we first review the development of large AI models in recent years.Then,from the perspective of methods,we introduce important theories and technologies of large decision models,such as model architecture and model adaptation.Next,from the perspective of applications,we introduce the cutting-edge applications of large decision models in various fields,such as autonomous driving and knowledge decision-making.Finally,we discuss existing challenges,such as security issues,decision bias and hallucination phenomenon as well as future prospects,from both technology development and domain applications.We hope this review paper can help researchers understand the important progress of intelligent decision-making driven by large AI models.
基金supported by grants from the National Key Research and Development Program of China(Grant No.2024YFC3405303)Beijing Natural Science Foundation(Grant No.7242281 and 7244427)Research and Development Fund of Peking University People’s Hospital(Grant No.RDZH2024-03 and RDEB2025-25).
文摘Objective:The current pathological diagnosis of lymph node metastasis is time-consuming,labor-intensive,and dependent on sectioning of paraffin blocks.Herein,in a prospective cohort of patients with breast cancer,we validated dynamic full-field optical coherence tomography(D-FFOCT),a virtual pathology tool integrating deep learning for nodal metastasis detection,and offering rapid and label-free histologic approximations of fresh tissues.Methods:In a prospective dual-center cohort of 155 patients with breast cancer,747 freshly bisected lymph node slides were obtained via D-FFOCT.Surgeons interpreted each slide with histopathology as the gold standard.A deep learning model was trained on 28,911 patches(corresponding to 590 slides)and tested on 7,736 patches(corresponding to 157 slides).The results were mapped to the slide level for potential intraoperative evaluation.Results:D-FFOCT strongly correlated with hematoxylin and eosin(H&E)-stained histological images.Surgeons achieved 97.10%specificity in nodal diagnosis with D-FFOCT.The performance of the artificial intelligence(AI)model was not inferior to that of human experts and had a sensitivity/specificity of 87.88%/91.94%and an area under the receiver operating characteristic curve of 0.899 at the slide level.The human–AI collaborative system reduced labor requirements by 75%and increased the specificity by 6.5%,to 98.39%.Conclusions:D-FFOCT has excellent potential as a tool for assessing lymph node metastatic status without tissue preparation or consumption.The integration of D-FFOCT with deep learning decreases labor demands and maintains high accuracy,thereby enabling streamlined nodal prediction independent of routine pathology procedures.
基金supported by National Natural Science Foundation of China (No. 62076251)sponsored by IMT-2020(5G) Promotion Group 5G+AI Work Group+3 种基金jointly sponsored by China Academy of Information and Communications TechnologyGuangdong OPPO Mobile Telecommunications Corp., Ltdvivo Mobile Communication Co., LtdHuawei Technologies Co., Ltd
文摘Artificial intelligence(AI)models are promising to improve the accuracy of wireless positioning systems,particularly in indoor environments where unpredictable radio propagation channel is a great challenge.Although great efforts have been made to explore the effectiveness of different AI models,it is still an open problem whether these models,trained with the data collected from all base stations(BSs),could work when some BSs are unavailable.In this paper,we make the first effort to enhance the generalization ability of AI wireless positioning model to adapt to the scenario where only partial BSs work.Particularly,a Siamese Network based Wireless Positioning Model(SNWPM)is proposed to predict the location of mobile user equipment from channel state information(CSI)collected from 5G BSs.Furthermore,a Feature Aware Attention Module(FAAM)is introduced to reinforce the capability of feature extraction from CSI data.Experiments are conducted on the 2022 Wireless Communication AI Competition(WAIC)dataset.The proposed SNWPM achieves decimeter-level positioning accuracy even if the data of partial BSs are unavailable.Compared with other AI models,the proposed SNWPM can reduce the positioning error by nearly 50%to more than 60%while using less parameters and lower computation resources.
文摘In a prior practice and policy article published in Healthcare Science,we introduced the deployed application of an artificial intelligence(AI)model to predict longer‐term inpatient readmissions to guide community care interventions for patients with complex conditions in the context of Singapore's Hospital to Home(H2H)program that has been operating since 2017.In this follow on practice and policy article,we further elaborate on Singapore's H2H program and care model,and its supporting AI model for multiple readmission prediction,in the following ways:(1)by providing updates on the AI and supporting information systems,(2)by reporting on customer engagement and related service delivery outcomes including staff‐related time savings and patient benefits in terms of bed days saved,(3)by sharing lessons learned with respect to(i)analytics challenges encountered due to the high degree of heterogeneity and resulting variability of the data set associated with the population of program participants,(ii)balancing competing needs for simpler and stable predictive models versus continuing to further enhance models and add yet more predictive variables,and(iii)the complications of continuing to make model changes when the AI part of the system is highly interlinked with supporting clinical information systems,(4)by highlighting how this H2H effort supported broader Covid‐19 response efforts across Singapore's public healthcare system,and finally(5)by commenting on how the experiences and related capabilities acquired from running this H2H program and related community care model and supporting AI prediction model are expected to contribute to the next wave of Singapore's public healthcare efforts from 2023 onwards.For the convenience of the reader,some content that introduces the H2H program and the multiple readmissions AI prediction model that previously appeared in the prior Healthcare Science publication is repeated at the beginning of this article.
基金supported in part by NSFC under Grant 62422407in part by RGC under Grant 26204424in part by ACCESS–AI Chip Center for Emerging Smart Systems, sponsored by the Inno HK initiative of the Innovation and Technology Commission of the Hong Kong Special Administrative Region Government
文摘Robotic computing systems play an important role in enabling intelligent robotic tasks through intelligent algo-rithms and supporting hardware.In recent years,the evolution of robotic algorithms indicates a roadmap from traditional robotics to hierarchical and end-to-end models.This algorithmic advancement poses a critical challenge in achieving balanced system-wide performance.Therefore,algorithm-hardware co-design has emerged as the primary methodology,which ana-lyzes algorithm behaviors on hardware to identify common computational properties.These properties can motivate algo-rithm optimization to reduce computational complexity and hardware innovation from architecture to circuit for high performance and high energy efficiency.We then reviewed recent works on robotic and embodied AI algorithms and computing hard-ware to demonstrate this algorithm-hardware co-design methodology.In the end,we discuss future research opportunities by answering two questions:(1)how to adapt the computing platforms to the rapid evolution of embodied AI algorithms,and(2)how to transform the potential of emerging hardware innovations into end-to-end inference improvements.
文摘By comparing price plans offered by several retail energy firms,end users with smart meters and controllers may optimize their energy use cost portfolios,due to the growth of deregulated retail power markets.To help smart grid end-users decrease power payment and usage unhappiness,this article suggests a decision system based on reinforcement learning to aid with electricity price plan selection.An enhanced state-based Markov decision process(MDP)without transition probabilities simulates the decision issue.A Kernel approximate-integrated batch Q-learning approach is used to tackle the given issue.Several adjustments to the sampling and data representation are made to increase the computational and prediction performance.Using a continuous high-dimensional state space,the suggested approach can uncover the underlying characteristics of time-varying pricing schemes.Without knowing anything regarding the market environment in advance,the best decision-making policy may be learned via case studies that use data from actual historical price plans.Experiments show that the suggested decision approach may reduce cost and energy usage dissatisfaction by using user data to build an accurate prediction strategy.In this research,we look at how smart city energy planners rely on precise load forecasts.It presents a hybrid method that extracts associated characteristics to improve accuracy in residential power consumption forecasts using machine learning(ML).It is possible to measure the precision of forecasts with the use of loss functions with the RMSE.This research presents a methodology for estimating smart home energy usage in response to the growing interest in explainable artificial intelligence(XAI).Using Shapley Additive explanations(SHAP)approaches,this strategy makes it easy for consumers to comprehend their energy use trends.To predict future energy use,the study employs gradient boosting in conjunction with long short-term memory neural networks.
文摘From globally popular video game Black Myth:Wukong,which has garnered a dedicated player base around the world,to DeepSeek,an artificial intelligence(AI)model developed at an impressively low cost that rivals U.S.company OpenAI’s ChatGPT,and the perfectly synchronized robotic ensemble performing with precision at this year’s China Central Television Spring Festival Gala,a Chinese New Year’s Eve extravaganza that aired on January 28-these big tech breakthroughs have risen to prominence one after another,generating massive buzz.
文摘In recent years,the rapid advancement of artificial intelligence(AI)has fostered deep integration between large AI models and robotic technology.Robots such as robotic dogs capable of carrying heavy loads on mountainous terrain or performing waste disposal tasks and humanoid robots that can execute high-precision component installations have gradually reached the public eye,raising expectations for embodied intelligent robots.
基金supported by the“Regional Innovation System&Education(RISE)”through the Seoul RISE Center,funded by the Ministry of Education(MOE)the Seoul Metropolitan Government(2025-RISE-01-018-04)supported by the Korea Digital Forensic Center.
文摘Time series anomaly detection is critical in domains such as manufacturing,finance,and cybersecurity.Recent generative AI models,particularly Transformer-and Autoencoder-based architectures,show strong accuracy but their robustness under noisy conditions is less understood.This study evaluates three representative models—AnomalyTransformer,TranAD,and USAD—on the Server Machine Dataset(SMD)and cross-domain benchmarks including the SoilMoisture Active Passive(SMAP)dataset,theMars Science Laboratory(MSL)dataset,and the Secure Water Treatment(SWaT)testbed.Seven noise settings(five canonical,two mixed)at multiple intensities are tested under fixed clean-data training,with variations in window,stride,and thresholding.Results reveal distinct robustness profiles:AnomalyTransformermaintains recall but loses precision under abrupt noise,TranAD balances sensitivity yet is vulnerable to structured anomalies,and USAD resists Gaussian perturbations but collapses under block anomalies.Quantitatively,F1 drops 60%–70%on noisy SMD,with severe collapse in SWaT(F1≤0.10,Drop up to 84%)but relative stability on SMAP/MSL(Drop within±10%).Overall,generative models exhibit complementary robustness patterns,highlighting noise-type dependent vulnerabilities and providing practical guidance for robust deployment.
文摘With the exponential growth in the complexity of Internet architectures and the widespread adoption of cloud-native service technologies,traditional operation and maintenance(Artificial Intelligence for IT Operations,AIOps)models-largely reliant on the paradigm of"expert rules+fixed scripts"-have become increasingly passive and inefficient when confronted with unknown faults and massive volumes of alerts.This study focuses on the application of large-scale AI model-based intelligent agents across the full lifecycle of Internet fault handling,aiming to construct autonomous O&M agents endowed with capabilities of perception,decision-making,and execution.The paper first analyzes the core challenges in current fault management:alert storms leading to missed and false incident reports,cross-system data silos hindering root cause localization,and heavy reliance on expert experience in manual troubleshooting,resulting in delayed response times.On this basis,a hierarchical solution architecture based on large-model agents is proposed,comprising a multi-source data perception layer,a fault reasoning and decision-making layer,and an automated execution layer[1].By integrating Retrieval-Augmented Generation(RAG)techniques with an O&M knowledge base,the proposed approach equips intelligent agents with the ability to interpret topology metrics,log semantics,and change events.Furthermore,the introduction of chain-of-thought reasoning and reflection mechanisms enables the agents to simulate expert diagnostic pathways,thereby achieving millisecond-level anomaly detection and minute-level root cause identification.
文摘Soil permeability is a critical parameter that dictates the movement of water through soil,and it impacts processes such as seepage,erosion,slope stability,foundation design,groundwater contamination,and various engineering applications.This study investigates the permeability of soil amended with waste foundry sand(WFS)at a replacement level of 10%.Permeability measurements are conducted for three distinct relative densities,spanning from 65% to 85%.The dataset compiled from these measurements is employed to develop ensemble artificial intelligence(AI)models.Specifically,four regressor AI models are considered:Nearest Neighbor(NNR),Decision Tree(DTR),Random Forest(RFR)and Support Vector Machine(SVR).These models are enhanced with four distinct base learners:Gradient Boosting(GB),Stacking Regressor(SR),AdaBoost Regressor(ADR),and XGBoost(XGB).The input parameters include fraction of base sand(BS),fraction of waste foundry sand(WFS),relative density(RD),duration of flow(T),quantity of flow(Q)and permeability(k),totalling 165 data points.Through comparative analysis,the Gradient Boost with Decision Tree(GB-DTR)model is found to be best-performed model,with R2=0.9919.Sensitivity analysis reveals that Q is the most influential input parameter in predicting soil permeability.
文摘With the rapid progress of AI technology,AI big models with transformer architecture as the core has made great progress in natural language processing,computer vision and other fields.Smart grid is a modern power system integrated with advanced information,communication and control technology.The complexity and variability of the system and the massive reference data provide application scenarios for the application of AI large model.This paper systematically expounds the key technologies matching with AI large model and its adaptability to the core links of smart grid,and focuses on the role of AI large model in smart grid construction,such as new energy grid connection,equipment management,grid topology optimization and dispatching decision,such as specific application modes and cases in load forecasting,real-time dispatching and multi-energy complementarity.At the same time,this paper deeply analyzes the key challenges in data,technology,engineering and security faced by the application of AI large model in various fields of power,and puts forward the corresponding optimal solutions.Finally,combined with typical cases,the future development direction of the integration of digital twins,generative AI and other technologies is conceived,which provides a theoretical reference and practical path for promoting the autonomous and efficient development of smart grid.
基金supported by the National Natural Science Foundation of China under grant nos.52277222,52406256,52177217the Shuimu Tsinghua Scholar Program(grant no.2022SM146)an Artificial Intelligence for Research Paradigm Reform Enabling Discipline Leapfrog Program Project Funding Grant.
文摘Rechargeable batteries are pivotal for achieving carbon neutrality and enabling the renewable energy transition.Their advancement requires inno-vations at micro(materials),device(manufacturing),and system(control and optimization)levels.However,traditional trial-and-error approaches are inadequate for modern scientific demands.As a transformative artificial intelligence(AI)technology,large language models(LLMs)deliver powerful semantic understanding and reasoning capabilities,driving a paradigm shift in battery research to address multilevel innovation needs.Neverthe-less,this field still faces dual challenges:ambiguous technical roadmaps and fragmented progress in stage-specific achievements.This review sys-tematically consolidates recent advances in applying LLMs to battery research,distilling core findings across four critical domains:knowledge integration,materials discovery,manufacturing processes,and system management.To address key bottlenecks—including limited model inter-pretability,inadequate alignment with electrochemical mechanisms,and real-world data adaptation challenges—we propose structured frameworks for deep integration of battery research and LLMs,alongside defined future technical pathways.These frameworks bridge fundamental battery science with AI-driven innovation paradigms to facilitate groundbreaking advances in next-generation battery technologies.
基金supported by the National Natural Science Foundation of China(61771154)the Fundamental Research Funds for the Central Universities(3072022CF0601)supported by Key Laboratory of Advanced Marine Communication and Information Technology,Ministry of Industry and Information Technology,Harbin Engineering University,Harbin,China.
文摘As modern communication technology advances apace,the digital communication signals identification plays an important role in cognitive radio networks,the communication monitoring and management systems.AI has become a promising solution to this problem due to its powerful modeling capability,which has become a consensus in academia and industry.However,because of the data-dependence and inexplicability of AI models and the openness of electromagnetic space,the physical layer digital communication signals identification model is threatened by adversarial attacks.Adversarial examples pose a common threat to AI models,where well-designed and slight perturbations added to input data can cause wrong results.Therefore,the security of AI models for the digital communication signals identification is the premise of its efficient and credible applications.In this paper,we first launch adversarial attacks on the end-to-end AI model for automatic modulation classifi-cation,and then we explain and present three defense mechanisms based on the adversarial principle.Next we present more detailed adversarial indicators to evaluate attack and defense behavior.Finally,a demonstration verification system is developed to show that the adversarial attack is a real threat to the digital communication signals identification model,which should be paid more attention in future research.
文摘Generating carbon credits in rural and wetland lagoon environments is important for the economic and social survival of the same.There are many methodologies to study and certificate the Carbon Sink such as the ISO 14064,VCS VERRA,UNI-BNEUTRAL,GOLD STANDARD and others.Many methods done before 2018 are obsolete since research has developed greatly in recent years.The methods are all different,but they share a continuous and real monitoring of the environment to ensure a true CCS(Carbon Capture and Storage)action.In the case of absence of monitoring,the method uses a system of provision of carbon credits called“buffer”.This system allows maintaining a credit-generating activity even in the presence of important anomalies due to adverse weather events.This research shows the complex analytic web of the different sensors in a continuous environmental monitoring system via GSM(Global System for Mobile)Communication and IoT(Internet of Things).By 2011,a monitoring network was installed in the wetland environments of Northern Italy Venetian Lagoon(UNESCO heritage)and used to understand and validate,the CCS action.Thingspeak cloud platform is used to collect data and is used to send alert to the user if the biological sink is reversed to emission.The obtained large dataset was used to prepare a AI(Artificial Intelligence)model“CCS wetland forecast”by Google COLAB.This model can fit the trend to avoid the direct and spot chemical field analysis and demonstrate the real efficacy of the model chosen.This network is now implemented by the Italian national method UNI PdR 99:2021 BNeutral generation of carbon credits.
