The development of communication technology will promote the application of Internet of Things,and Beyond 5G will become a new technology promoter.At the same time,Beyond 5G will become one of the important supports f...The development of communication technology will promote the application of Internet of Things,and Beyond 5G will become a new technology promoter.At the same time,Beyond 5G will become one of the important supports for the development of edge computing technology.This paper proposes a communication task allocation algorithm based on deep reinforcement learning for vehicle-to-pedestrian communication scenarios in edge computing.Through trial and error learning of agent,the optimal spectrum and power can be determined for transmission without global information,so as to balance the communication between vehicle-to-pedestrian and vehicle-to-infrastructure.The results show that the agent can effectively improve vehicle-to-infrastructure communication rate as well as meeting the delay constraints on the vehicle-to-pedestrian link.展开更多
Directly grasping the tightly stacked objects may cause collisions and result in failures,degenerating the functionality of robotic arms.Inspired by the observation that first pushing objects to a state of mutual sepa...Directly grasping the tightly stacked objects may cause collisions and result in failures,degenerating the functionality of robotic arms.Inspired by the observation that first pushing objects to a state of mutual separation and then grasping them individually can effectively increase the success rate,we devise a novel deep Q-learning framework to achieve collaborative pushing and grasping.Specifically,an efficient non-maximum suppression policy(PolicyNMS)is proposed to dynamically evaluate pushing and grasping actions by enforcing a suppression constraint on unreasonable actions.Moreover,a novel data-driven pushing reward network called PR-Net is designed to effectively assess the degree of separation or aggregation between objects.To benchmark the proposed method,we establish a dataset containing common household items dataset(CHID)in both simulation and real scenarios.Although trained using simulation data only,experiment results validate that our method generalizes well to real scenarios and achieves a 97%grasp success rate at a fast speed for object separation in the real-world environment.展开更多
The Chinese economy achieved a 5.0%growth in 2024,significantly exceeding expectations.In the third quarter of 2024,China's GDP growth had declined to 4.6%,while deflationary risks increased substantially.In last ...The Chinese economy achieved a 5.0%growth in 2024,significantly exceeding expectations.In the third quarter of 2024,China's GDP growth had declined to 4.6%,while deflationary risks increased substantially.In last September,the top policymaking body called for“strengthening counter-cyclical adjustments”,leading to the implementation of a new round of stimulus measures.These policies aimed at stabilizing the stock market,real estate sector,and overall economic growth,including an ambitious 12 trillion yuan debt swap program for local governments over the coming years.These concerted efforts successfully reversed the downward trend and contributed to economic stabilization in the fourth quarter of 2024.展开更多
Large language models(LLMs)exhibit remarkable capabilities in various natural language processing tasks,such as machine translation.However,the large number of LLM parameters incurs significant costs during inference....Large language models(LLMs)exhibit remarkable capabilities in various natural language processing tasks,such as machine translation.However,the large number of LLM parameters incurs significant costs during inference.Previous studies have attempted to train translation-tailored LLMs with moderately sized models by fine-tuning them on the translation data.Nevertheless,when performing translations in zero-shot directions that are absent from the fine-tuning data,the problem of ignoring instructions and thus producing translations in the wrong language(i.e.,the off-target translation issue)remains unresolved.In this work,we design a twostage fine-tuning algorithm to improve the instruction-following ability of translation-tailored LLMs,particularly for maintaining accurate translation directions.We first fine-tune LLMs on the translation data to elicit basic translation capabilities.At the second stage,we construct instruction-conficting samples by randomly replacing the instructions with the incorrect ones.Then,we introduce an extra unlikelihood loss to reduce the probability assigned to those samples.Experiments on two benchmarks using the LLaMA 2 and LLaMA 3 models,spanning 16 zero-shot directions,demonstrate that,compared to the competitive baseline translation-finetuned LLaMA,our method could effectively reduce the off-target translation ratio(up to-62.4 percentage points),thus improving translation quality(up to+9.7 bilingual evaluation understudy).Analysis shows that our method can preserve the model's performance on other tasks,such as supervised translation and general tasks.Code is released at https://github.com/alphadl/LanguageAware_Tuning.展开更多
Scene-based recommendation has proven its usefulness in E-commerce,by recommending commodities based on a given scene.However,scenes are typically unknown in advance,which necessitates scene discovery for E-commerce.I...Scene-based recommendation has proven its usefulness in E-commerce,by recommending commodities based on a given scene.However,scenes are typically unknown in advance,which necessitates scene discovery for E-commerce.In this article,we study scene discovery for E-commerce systems.We first formalize a scene as a set of commodity cate-gories that occur simultaneously and frequently in real-world situations,and model an E-commerce platform as a heteroge-neous information network(HIN),whose nodes and links represent different types of objects and different types of rela-tionships between objects,respectively.We then formulate the scene mining problem for E-commerce as an unsupervised learning problem that finds the overlapping clusters of commodity categories in the HIN.To solve the problem,we pro-pose a non-negative matrix factorization based method SMEC(Scene Mining for E-Commerce),and theoretically prove its convergence.Using six real-world E-commerce datasets,we finally conduct an extensive experimental study to evaluate SMEC against 13 other methods,and show that SMEC consistently outperforms its competitors with regard to various evaluation measures.展开更多
Since 2020,the COVID-19 pandemic has spread globally,causing severe damage to national economies,halting trade and personnel exchanges,and pushing industrial chains to the brink of collapse.In response,the US and Euro...Since 2020,the COVID-19 pandemic has spread globally,causing severe damage to national economies,halting trade and personnel exchanges,and pushing industrial chains to the brink of collapse.In response,the US and European economies implemented unprecedented fiscal and monetary expansionary policies to avert economic recession,but this also led to a sharp rise in inflation.During this period,the outbreak of the Ukraine crisis further exacerbated inflation and more importantly,triggered an evolution in the global geopolitical landscape.展开更多
Recently,global capital markets have been swept up in sharp.volatility,with stock and exchange rate markets going through roller-coaster rides.On August 5,the Japanese stock market fell.The Nikkei 225 index plunged 12...Recently,global capital markets have been swept up in sharp.volatility,with stock and exchange rate markets going through roller-coaster rides.On August 5,the Japanese stock market fell.The Nikkei 225 index plunged 12.4%,triggering circuit breakers twice during the session and wiping out all gains made earlier this year.The Korea Composite Stock Price Index(KOSPI)fell by 8.8%and triggered the circuit breaker mechanism.Meanwhile,the three major US stock indices also declined,though they were contained in a narrower range of 2-3%.But the next day,global stock markets rebounded strongly.The Nikkei 225 index surged by 10.2%,KOSPI rose by 3.3%,and all three major US stock indices saw an upswing.展开更多
With the increasing prevalence of Android software,protecting it against malicious threats has become a critical concern.Traditional malware detection methods,tailored for static environments,often fail to adapt to ev...With the increasing prevalence of Android software,protecting it against malicious threats has become a critical concern.Traditional malware detection methods,tailored for static environments,often fail to adapt to evolving threats in dynamic environments.To address the challenge of detecting evolving malware,we introduce DMDroid,a novel multi-modal fusion-based framework for malware analysis and detection.DMDroid leverages an array of feature extraction technologies and advanced deep learning models to analyze data,enhanced by a multi-head attention mechanism.This mechanism optimizes the integration of diverse static features from graphbased and image-based modalities,including permissions,API calls,opcodes,and bytecode sequences,prioritizing critical features to effectively detect new and evolving malware threats.We evaluate DMDroid in various realistic environments.Experiments show that compared to Bai,Drebin,and MaMa-pkg detector,DMDroid can improve the detection accuracy by 117.56%,122.11%,and 119.47%,respectively.Compared to an unimodal approach,DMDroid can enhance the accuracy,macro-averaged F1 score,and weighted-averaged F1 score by 143.25%,75.84%and 279.22%.The prototype can help to improve the quality and security of Android malware analysis and detection.展开更多
Today’s supply chain is becoming complex and fragile.Hence,supply chain managers need to create and unlock the value of the smart supply chain.A smart supply chain requires connectivity,visibility,and agility,and it ...Today’s supply chain is becoming complex and fragile.Hence,supply chain managers need to create and unlock the value of the smart supply chain.A smart supply chain requires connectivity,visibility,and agility,and it needs be integrated and intelligent.The digital twin(DT)concept satisfies these requirements.Therefore,we propose creating a DT-driven supply chain(DTSC)as an innovative and integrated solution for the smart supply chain.We provide background information to explain the DT concept and to demonstrate the method for building a DTSC by using the DT concept.We discuss three research opportunities in building a DTSC,including supply chain modeling,real-time supply chain optimization,and data usage in supply chain collaboration.Finally,we highlight a motivating case from JD.COM,China’s largest retailer by revenue,in applying the DTSC platform to address supply chain network reconfiguration challenges during the COVID-19 pandemic.展开更多
A virtual cosmetics try-on system provides a realistic try-on experience for consumers and helps them efficiently choose suitable cosmetics.In this article,we propose a real-time augmented reality virtual cosmetics tr...A virtual cosmetics try-on system provides a realistic try-on experience for consumers and helps them efficiently choose suitable cosmetics.In this article,we propose a real-time augmented reality virtual cosmetics try-on system for smartphones(ARCosmetics),taking speed,accuracy,and stability into consideration at each step to ensure a better user experience.A novel and very fast face tracking method utilizes the face detection box and the average position of facial landmarks to estimate the faces in continuous frames.A dynamic weight Wing loss is introduced to assign a dynamic weight to every landmark by the estimated error during training.It balances the attention between small,medium,and large range error and thus increases the accuracy and robustness.We also designed a weighted average method to utilize the information of the adjacent frame for landmark refinement,guaranteeing the stability of the generated landmarks.Extensive experiments conducted on a large 106-point facial landmark dataset and the 300-VW dataset demonstrate the superior performance of the proposed method compared to other state-of-the-art methods.We also conducted user satisfaction studies further to verify the efficiency and effectiveness of our ARCosmetics system.展开更多
The recovery of consumption directly matters economic development.In 2022,final consumption contributed 32.8%to the GDP,well below the average level of about 65% in previous years,being the main reason for the slowdow...The recovery of consumption directly matters economic development.In 2022,final consumption contributed 32.8%to the GDP,well below the average level of about 65% in previous years,being the main reason for the slowdown in economic growth.The repair of travel led to a rebound in services and contact-based service consumption in 2023,driving the growth rate of total retail sales to 3.5% in January and February.Total retail sales from March to May were also up compared to last year,this was mainly due to a lower base in 2022.展开更多
In 2022,China's total retail sales fell 0.2%year on year,the second year of negative growth since 1969(the last time was 2020 when COVID-19 broke out),while household savings deposits rose sharply by 17.8 trillion...In 2022,China's total retail sales fell 0.2%year on year,the second year of negative growth since 1969(the last time was 2020 when COVID-19 broke out),while household savings deposits rose sharply by 17.8 trillion yuan,almost 8 trillion yuan more than in 2021,a record high.Politeymakers want a targeted solution to insuffcient demand."Using consumption as a lever to accelerate economie reeovery"has sparked a heated debate on whether high savings can drive consumption reeovery.展开更多
Leveraging algorithms to provide performance feedback to employees has become widespread in organizations.Algorithm-generated feedback is quite different from human’s feedback in feedback form and employees’percepti...Leveraging algorithms to provide performance feedback to employees has become widespread in organizations.Algorithm-generated feedback is quite different from human’s feedback in feedback form and employees’perceptions,so it is hard to directly predict the effect of algorithm-generated feedbacks.Despite the widespread use of algorithm-generated feedback in workplace,there is scant empirical evidence revealing its impacts.To address this gap,we empirically examine the effects of the implementation of an algorithm-generated feedback system through a field experiment conducted in the logistics industry.The results indicated that the algorithm-generated feedback significantly reduces customer complaints by about 20%.Additionally,employees with less work experience or lower workloads benefit from algorithm-generated feedback more.This work offers empirical evidence on the business value of algorithm-generated feedback and highlights the importance of employee characteristics in understanding and managing the effects of algorithmic supervision in the workplace.展开更多
基金supported by National Natural Science Foundation of China(No.61871283)the Foundation of Pre-Research on Equipment of China(No.61400010304)Major Civil-Military Integration Project in Tianjin,China(No.18ZXJMTG00170).
文摘The development of communication technology will promote the application of Internet of Things,and Beyond 5G will become a new technology promoter.At the same time,Beyond 5G will become one of the important supports for the development of edge computing technology.This paper proposes a communication task allocation algorithm based on deep reinforcement learning for vehicle-to-pedestrian communication scenarios in edge computing.Through trial and error learning of agent,the optimal spectrum and power can be determined for transmission without global information,so as to balance the communication between vehicle-to-pedestrian and vehicle-to-infrastructure.The results show that the agent can effectively improve vehicle-to-infrastructure communication rate as well as meeting the delay constraints on the vehicle-to-pedestrian link.
基金This work was supported by the National Natural Science Foundation of China(61873077,61806062)Zhejiang Provincial Major Research and Development Project of China(2020C01110)Zhejiang Provincial Key Laboratory of Equipment Electronics.
文摘Directly grasping the tightly stacked objects may cause collisions and result in failures,degenerating the functionality of robotic arms.Inspired by the observation that first pushing objects to a state of mutual separation and then grasping them individually can effectively increase the success rate,we devise a novel deep Q-learning framework to achieve collaborative pushing and grasping.Specifically,an efficient non-maximum suppression policy(PolicyNMS)is proposed to dynamically evaluate pushing and grasping actions by enforcing a suppression constraint on unreasonable actions.Moreover,a novel data-driven pushing reward network called PR-Net is designed to effectively assess the degree of separation or aggregation between objects.To benchmark the proposed method,we establish a dataset containing common household items dataset(CHID)in both simulation and real scenarios.Although trained using simulation data only,experiment results validate that our method generalizes well to real scenarios and achieves a 97%grasp success rate at a fast speed for object separation in the real-world environment.
文摘The Chinese economy achieved a 5.0%growth in 2024,significantly exceeding expectations.In the third quarter of 2024,China's GDP growth had declined to 4.6%,while deflationary risks increased substantially.In last September,the top policymaking body called for“strengthening counter-cyclical adjustments”,leading to the implementation of a new round of stimulus measures.These policies aimed at stabilizing the stock market,real estate sector,and overall economic growth,including an ambitious 12 trillion yuan debt swap program for local governments over the coming years.These concerted efforts successfully reversed the downward trend and contributed to economic stabilization in the fourth quarter of 2024.
基金Project supported by the National Natural Science Foundation of China(No.62372468)the Shandong Natural Science Foundation(No.ZR2023MF008)+1 种基金the Major Basic Research Projects in Shandong Province(No.ZR2023ZD32)the Qingdao Natural Science Foundation(No.23-2-1-161-zyyd-jch)。
文摘Large language models(LLMs)exhibit remarkable capabilities in various natural language processing tasks,such as machine translation.However,the large number of LLM parameters incurs significant costs during inference.Previous studies have attempted to train translation-tailored LLMs with moderately sized models by fine-tuning them on the translation data.Nevertheless,when performing translations in zero-shot directions that are absent from the fine-tuning data,the problem of ignoring instructions and thus producing translations in the wrong language(i.e.,the off-target translation issue)remains unresolved.In this work,we design a twostage fine-tuning algorithm to improve the instruction-following ability of translation-tailored LLMs,particularly for maintaining accurate translation directions.We first fine-tune LLMs on the translation data to elicit basic translation capabilities.At the second stage,we construct instruction-conficting samples by randomly replacing the instructions with the incorrect ones.Then,we introduce an extra unlikelihood loss to reduce the probability assigned to those samples.Experiments on two benchmarks using the LLaMA 2 and LLaMA 3 models,spanning 16 zero-shot directions,demonstrate that,compared to the competitive baseline translation-finetuned LLaMA,our method could effectively reduce the off-target translation ratio(up to-62.4 percentage points),thus improving translation quality(up to+9.7 bilingual evaluation understudy).Analysis shows that our method can preserve the model's performance on other tasks,such as supervised translation and general tasks.Code is released at https://github.com/alphadl/LanguageAware_Tuning.
基金The work was supported by the National Key Research and Development Program of China under Grant No.2018AAA0102301the National Natural Science Foundation of China under Grant No.61925203.
文摘Scene-based recommendation has proven its usefulness in E-commerce,by recommending commodities based on a given scene.However,scenes are typically unknown in advance,which necessitates scene discovery for E-commerce.In this article,we study scene discovery for E-commerce systems.We first formalize a scene as a set of commodity cate-gories that occur simultaneously and frequently in real-world situations,and model an E-commerce platform as a heteroge-neous information network(HIN),whose nodes and links represent different types of objects and different types of rela-tionships between objects,respectively.We then formulate the scene mining problem for E-commerce as an unsupervised learning problem that finds the overlapping clusters of commodity categories in the HIN.To solve the problem,we pro-pose a non-negative matrix factorization based method SMEC(Scene Mining for E-Commerce),and theoretically prove its convergence.Using six real-world E-commerce datasets,we finally conduct an extensive experimental study to evaluate SMEC against 13 other methods,and show that SMEC consistently outperforms its competitors with regard to various evaluation measures.
文摘Since 2020,the COVID-19 pandemic has spread globally,causing severe damage to national economies,halting trade and personnel exchanges,and pushing industrial chains to the brink of collapse.In response,the US and European economies implemented unprecedented fiscal and monetary expansionary policies to avert economic recession,but this also led to a sharp rise in inflation.During this period,the outbreak of the Ukraine crisis further exacerbated inflation and more importantly,triggered an evolution in the global geopolitical landscape.
文摘Recently,global capital markets have been swept up in sharp.volatility,with stock and exchange rate markets going through roller-coaster rides.On August 5,the Japanese stock market fell.The Nikkei 225 index plunged 12.4%,triggering circuit breakers twice during the session and wiping out all gains made earlier this year.The Korea Composite Stock Price Index(KOSPI)fell by 8.8%and triggered the circuit breaker mechanism.Meanwhile,the three major US stock indices also declined,though they were contained in a narrower range of 2-3%.But the next day,global stock markets rebounded strongly.The Nikkei 225 index surged by 10.2%,KOSPI rose by 3.3%,and all three major US stock indices saw an upswing.
基金supported by a sub-project of the National Key Research and Development Program of the Ministry of Science and Technology,with grant number 2022YFB4501700
文摘With the increasing prevalence of Android software,protecting it against malicious threats has become a critical concern.Traditional malware detection methods,tailored for static environments,often fail to adapt to evolving threats in dynamic environments.To address the challenge of detecting evolving malware,we introduce DMDroid,a novel multi-modal fusion-based framework for malware analysis and detection.DMDroid leverages an array of feature extraction technologies and advanced deep learning models to analyze data,enhanced by a multi-head attention mechanism.This mechanism optimizes the integration of diverse static features from graphbased and image-based modalities,including permissions,API calls,opcodes,and bytecode sequences,prioritizing critical features to effectively detect new and evolving malware threats.We evaluate DMDroid in various realistic environments.Experiments show that compared to Bai,Drebin,and MaMa-pkg detector,DMDroid can improve the detection accuracy by 117.56%,122.11%,and 119.47%,respectively.Compared to an unimodal approach,DMDroid can enhance the accuracy,macro-averaged F1 score,and weighted-averaged F1 score by 143.25%,75.84%and 279.22%.The prototype can help to improve the quality and security of Android malware analysis and detection.
基金The authors are grateful for the financial support from the National Key R&D Program of China(Grant No.2018YFB1700600).
文摘Today’s supply chain is becoming complex and fragile.Hence,supply chain managers need to create and unlock the value of the smart supply chain.A smart supply chain requires connectivity,visibility,and agility,and it needs be integrated and intelligent.The digital twin(DT)concept satisfies these requirements.Therefore,we propose creating a DT-driven supply chain(DTSC)as an innovative and integrated solution for the smart supply chain.We provide background information to explain the DT concept and to demonstrate the method for building a DTSC by using the DT concept.We discuss three research opportunities in building a DTSC,including supply chain modeling,real-time supply chain optimization,and data usage in supply chain collaboration.Finally,we highlight a motivating case from JD.COM,China’s largest retailer by revenue,in applying the DTSC platform to address supply chain network reconfiguration challenges during the COVID-19 pandemic.
基金supported in part by the National Key R&D Program of China(2021ZD0140407)in part by the National Natural Science Foundation of China(Grant No.U21A20523).
文摘A virtual cosmetics try-on system provides a realistic try-on experience for consumers and helps them efficiently choose suitable cosmetics.In this article,we propose a real-time augmented reality virtual cosmetics try-on system for smartphones(ARCosmetics),taking speed,accuracy,and stability into consideration at each step to ensure a better user experience.A novel and very fast face tracking method utilizes the face detection box and the average position of facial landmarks to estimate the faces in continuous frames.A dynamic weight Wing loss is introduced to assign a dynamic weight to every landmark by the estimated error during training.It balances the attention between small,medium,and large range error and thus increases the accuracy and robustness.We also designed a weighted average method to utilize the information of the adjacent frame for landmark refinement,guaranteeing the stability of the generated landmarks.Extensive experiments conducted on a large 106-point facial landmark dataset and the 300-VW dataset demonstrate the superior performance of the proposed method compared to other state-of-the-art methods.We also conducted user satisfaction studies further to verify the efficiency and effectiveness of our ARCosmetics system.
文摘The recovery of consumption directly matters economic development.In 2022,final consumption contributed 32.8%to the GDP,well below the average level of about 65% in previous years,being the main reason for the slowdown in economic growth.The repair of travel led to a rebound in services and contact-based service consumption in 2023,driving the growth rate of total retail sales to 3.5% in January and February.Total retail sales from March to May were also up compared to last year,this was mainly due to a lower base in 2022.
文摘In 2022,China's total retail sales fell 0.2%year on year,the second year of negative growth since 1969(the last time was 2020 when COVID-19 broke out),while household savings deposits rose sharply by 17.8 trillion yuan,almost 8 trillion yuan more than in 2021,a record high.Politeymakers want a targeted solution to insuffcient demand."Using consumption as a lever to accelerate economie reeovery"has sparked a heated debate on whether high savings can drive consumption reeovery.
基金supported by the Fundamental Research Funds for the Central Universities,and the Research Funds of Renmin University of China(24XNKJ12)the National Natural Science Foundation of China(72201038).
文摘Leveraging algorithms to provide performance feedback to employees has become widespread in organizations.Algorithm-generated feedback is quite different from human’s feedback in feedback form and employees’perceptions,so it is hard to directly predict the effect of algorithm-generated feedbacks.Despite the widespread use of algorithm-generated feedback in workplace,there is scant empirical evidence revealing its impacts.To address this gap,we empirically examine the effects of the implementation of an algorithm-generated feedback system through a field experiment conducted in the logistics industry.The results indicated that the algorithm-generated feedback significantly reduces customer complaints by about 20%.Additionally,employees with less work experience or lower workloads benefit from algorithm-generated feedback more.This work offers empirical evidence on the business value of algorithm-generated feedback and highlights the importance of employee characteristics in understanding and managing the effects of algorithmic supervision in the workplace.
