In public goods games, punishments and rewards have been shown to be effective mechanisms for maintaining individualcooperation. However, punishments and rewards are costly to incentivize cooperation. Therefore, the g...In public goods games, punishments and rewards have been shown to be effective mechanisms for maintaining individualcooperation. However, punishments and rewards are costly to incentivize cooperation. Therefore, the generation ofcostly penalties and rewards has been a complex problem in promoting the development of cooperation. In real society,specialized institutions exist to punish evil people or reward good people by collecting taxes. We propose a strong altruisticpunishment or reward strategy in the public goods game through this phenomenon. Through theoretical analysis and numericalcalculation, we can get that tax-based strong altruistic punishment (reward) has more evolutionary advantages thantraditional strong altruistic punishment (reward) in maintaining cooperation and tax-based strong altruistic reward leads toa higher level of cooperation than tax-based strong altruistic punishment.展开更多
To explore the green development of automobile enterprises and promote the achievement of the“dual carbon”target,based on the bounded rationality assumptions,this study constructed a tripartite evolutionary game mod...To explore the green development of automobile enterprises and promote the achievement of the“dual carbon”target,based on the bounded rationality assumptions,this study constructed a tripartite evolutionary game model of gov-ernment,commercial banks,and automobile enterprises;introduced a dynamic reward and punishment mechanism;and analyzed the development process of the three parties’strategic behavior under the static and dynamic reward and punish-ment mechanism.Vensim PLE was used for numerical simulation analysis.Our results indicate that the system could not reach a stable state under the static reward and punishment mechanism.A dynamic reward and punishment mechanism can effectively improve the system stability and better fit real situations.Under the dynamic reward and punishment mechan-ism,an increase in the initial probabilities of the three parties can promote the system stability,and the government can im-plement effective supervision by adjusting the upper limit of the reward and punishment intensity.Finally,the implementa-tion of green credit by commercial banks plays a significant role in promoting the green development of automobile enter-prises.展开更多
By integrating deep neural networks with reinforcement learning,the Double Deep Q Network(DDQN)algorithm overcomes the limitations of Q-learning in handling continuous spaces and is widely applied in the path planning...By integrating deep neural networks with reinforcement learning,the Double Deep Q Network(DDQN)algorithm overcomes the limitations of Q-learning in handling continuous spaces and is widely applied in the path planning of mobile robots.However,the traditional DDQN algorithm suffers from sparse rewards and inefficient utilization of high-quality data.Targeting those problems,an improved DDQN algorithm based on average Q-value estimation and reward redistribution was proposed.First,to enhance the precision of the target Q-value,the average of multiple previously learned Q-values from the target Q network is used to replace the single Q-value from the current target Q network.Next,a reward redistribution mechanism is designed to overcome the sparse reward problem by adjusting the final reward of each action using the round reward from trajectory information.Additionally,a reward-prioritized experience selection method is introduced,which ranks experience samples according to reward values to ensure frequent utilization of high-quality data.Finally,simulation experiments are conducted to verify the effectiveness of the proposed algorithm in fixed-position scenario and random environments.The experimental results show that compared to the traditional DDQN algorithm,the proposed algorithm achieves shorter average running time,higher average return and fewer average steps.The performance of the proposed algorithm is improved by 11.43%in the fixed scenario and 8.33%in random environments.It not only plans economic and safe paths but also significantly improves efficiency and generalization in path planning,making it suitable for widespread application in autonomous navigation and industrial automation.展开更多
Cross-lingual image description,the task of generating image captions in a target language from images and descriptions in a source language,is addressed in this study through a novel approach that combines neural net...Cross-lingual image description,the task of generating image captions in a target language from images and descriptions in a source language,is addressed in this study through a novel approach that combines neural network models and semantic matching techniques.Experiments conducted on the Flickr8k and AraImg2k benchmark datasets,featuring images and descriptions in English and Arabic,showcase remarkable performance improvements over state-of-the-art methods.Our model,equipped with the Image&Cross-Language Semantic Matching module and the Target Language Domain Evaluation module,significantly enhances the semantic relevance of generated image descriptions.For English-to-Arabic and Arabic-to-English cross-language image descriptions,our approach achieves a CIDEr score for English and Arabic of 87.9%and 81.7%,respectively,emphasizing the substantial contributions of our methodology.Comparative analyses with previous works further affirm the superior performance of our approach,and visual results underscore that our model generates image captions that are both semantically accurate and stylistically consistent with the target language.In summary,this study advances the field of cross-lingual image description,offering an effective solution for generating image captions across languages,with the potential to impact multilingual communication and accessibility.Future research directions include expanding to more languages and incorporating diverse visual and textual data sources.展开更多
Autonomous umanned aerial vehicle(UAV) manipulation is necessary for the defense department to execute tactical missions given by commanders in the future unmanned battlefield. A large amount of research has been devo...Autonomous umanned aerial vehicle(UAV) manipulation is necessary for the defense department to execute tactical missions given by commanders in the future unmanned battlefield. A large amount of research has been devoted to improving the autonomous decision-making ability of UAV in an interactive environment, where finding the optimal maneuvering decisionmaking policy became one of the key issues for enabling the intelligence of UAV. In this paper, we propose a maneuvering decision-making algorithm for autonomous air-delivery based on deep reinforcement learning under the guidance of expert experience. Specifically, we refine the guidance towards area and guidance towards specific point tasks for the air-delivery process based on the traditional air-to-surface fire control methods.Moreover, we construct the UAV maneuvering decision-making model based on Markov decision processes(MDPs). Specifically, we present a reward shaping method for the guidance towards area and guidance towards specific point tasks using potential-based function and expert-guided advice. The proposed algorithm could accelerate the convergence of the maneuvering decision-making policy and increase the stability of the policy in terms of the output during the later stage of training process. The effectiveness of the proposed maneuvering decision-making policy is illustrated by the curves of training parameters and extensive experimental results for testing the trained policy.展开更多
Mobile adhoc networks have grown in prominence in recent years,and they are now utilized in a broader range of applications.The main challenges are related to routing techniques that are generally employed in them.Mob...Mobile adhoc networks have grown in prominence in recent years,and they are now utilized in a broader range of applications.The main challenges are related to routing techniques that are generally employed in them.Mobile Adhoc system management,on the other hand,requires further testing and improvements in terms of security.Traditional routing protocols,such as Adhoc On-Demand Distance Vector(AODV)and Dynamic Source Routing(DSR),employ the hop count to calculate the distance between two nodes.The main aim of this research work is to determine the optimum method for sending packets while also extending life time of the network.It is achieved by changing the residual energy of each network node.Also,in this paper,various algorithms for optimal routing based on parameters like energy,distance,mobility,and the pheromone value are proposed.Moreover,an approach based on a reward and penalty system is given in this paper to evaluate the efficiency of the proposed algorithms under the impact of parameters.The simulation results unveil that the reward penalty-based approach is quite effective for the selection of an optimal path for routing when the algorithms are implemented under the parameters of interest,which helps in achieving less packet drop and energy consumption of the nodes along with enhancing the network efficiency.展开更多
Goal-conditioned reinforcement learning(RL)is an interesting extension of the traditional RL framework,where the dynamic environment and reward sparsity can cause conventional learning algorithms to fail.Reward shapin...Goal-conditioned reinforcement learning(RL)is an interesting extension of the traditional RL framework,where the dynamic environment and reward sparsity can cause conventional learning algorithms to fail.Reward shaping is a practical approach to improving sample efficiency by embedding human domain knowledge into the learning process.Existing reward shaping methods for goal-conditioned RL are typically built on distance metrics with a linear and isotropic distribution,which may fail to provide sufficient information about the ever-changing environment with high complexity.This paper proposes a novel magnetic field-based reward shaping(MFRS)method for goal-conditioned RL tasks with dynamic target and obstacles.Inspired by the physical properties of magnets,we consider the target and obstacles as permanent magnets and establish the reward function according to the intensity values of the magnetic field generated by these magnets.The nonlinear and anisotropic distribution of the magnetic field intensity can provide more accessible and conducive information about the optimization landscape,thus introducing a more sophisticated magnetic reward compared to the distance-based setting.Further,we transform our magnetic reward to the form of potential-based reward shaping by learning a secondary potential function concurrently to ensure the optimal policy invariance of our method.Experiments results in both simulated and real-world robotic manipulation tasks demonstrate that MFRS outperforms relevant existing methods and effectively improves the sample efficiency of RL algorithms in goal-conditioned tasks with various dynamics of the target and obstacles.展开更多
As assessment outcomes provide students with a sense of accomplishment that is boosted by the reward system,learning becomes more effective.This research aims to determine the effects of reward system prior to assessm...As assessment outcomes provide students with a sense of accomplishment that is boosted by the reward system,learning becomes more effective.This research aims to determine the effects of reward system prior to assessment in Mathematics.Quasi-experimental research design was used to examine whether there was a significant difference between the use of reward system and students’level of performance in Mathematics.Through purposive sampling,the respondents of the study involve 80 Grade 9 students belonging to two sections from Gaudencio B.Lontok Memorial Integrated School.Based on similar demographics and pre-test results,control and study group were involved as participants of the study.Data were treated and analyzed accordingly using statistical treatments such as mean and t-test for independent variables.There was a significant finding revealing the advantage of using the reward system compare to the non-reward system in increasing students’level of performance in Mathematics.It is concluded that the use of reward system is effective in improving the assessment outcomes in Mathematics.It is recommended to use reward system for persistent assessment outcomes prior to assessment,to be a reflection of the intended outcomes in Mathematics.展开更多
Grassland ecological protection compensation and reward policy is the largest-scale investment concerning themost extensive areas since foundation of the PRC. It will be the long-term implementationpolicy for grasslan...Grassland ecological protection compensation and reward policy is the largest-scale investment concerning themost extensive areas since foundation of the PRC. It will be the long-term implementationpolicy for grassland ecological protection. In this study,based on macro-perspective, the policy effects ofgrasslandproductivity, ecological protection, animal husbandryoutput, pastoralists＇ income were ana- lyzed. The resultsshow that, afterimplementation of the policy, naturalgrass production and grasslandtheoretical stocking rateincreased. The averagenatural grasslandlivestockoverloading ratedecreased significantly, comprehensivenationalgrasslandvegetation coverageis increasing. Besides, adult cattleandbeef yield arefluctuated. Sheep head, adult sheep, sheep production, milk productionincreasedin varying degrees. The per capita netincomeof farmers and pastoralists, livestock income, the proportion oflivestockincomewere higher than those beforeimplementation of the policy.展开更多
The orbitofrontal cortex （OFC） is particularly important for the neural representation of reward value. Previous studies indicated that electroencephalogram （EEG） activity in the OFC was involved in drug administr...The orbitofrontal cortex （OFC） is particularly important for the neural representation of reward value. Previous studies indicated that electroencephalogram （EEG） activity in the OFC was involved in drug administration and withdrawal. The present study investigated EEG activity in the OFC in rats during the development of food reward and craving. Two environments were used separately for control and food-related EEG recordings. In the food-related environment rats were first trained to eat chocolate peanuts; then they either had no access to this food, but could see and smell it （craving trials）, or had free access to this food （reward trials）. The EEG in the left OFC was recorded during these trials. We showed that, in the food-related environment the EEG activity peaking in the delta band （2-4 Hz） was significantly correlated with the stimulus, increasing during food reward and decreasing during food craving when compared with that in the control environment. Our data suggests that EEG activity in the OFC can be altered by food reward; moreover, delta rhythm in this region could be used as an index monitoring changed signal underlying this reward.展开更多
Objective To investigate the co-effect of Demand-control-support （DCS） model and Effort-reward Imbalance （ERI） model on the risk estimation of depression in humans in comparison with the effects when they are used...Objective To investigate the co-effect of Demand-control-support （DCS） model and Effort-reward Imbalance （ERI） model on the risk estimation of depression in humans in comparison with the effects when they are used respectively. Methods A total of 3 632 males and 1 706 females from 13 factories and companies in Henan province were recruited in this cross-sectional study. Perceived job stress was evaluated with the Job Content Questionnaire and Effort-Reward Imbalance Questionnaire （Chinese version）. Depressive symptoms were assessed by using the Center for Epidemiological Studies Depression Scale （CES-D）. Results DC （demands/job control ratio） and ERI were shown to be independently associated with depressive symptoms. The outcome of low social support and overcommitment were similar. High DC and low social support （SS）, high ERI and high overcommitment, and high DC and high ERI posed greater risks of depressive symptoms than each of them did alone. ERI model and SS model seem to be effective in estimating the risk of depressive symptoms if they are used respectively. Conclusion The DC had better performance when it was used in combination with low SS. The effect on physical demands was better than on psychological demands. The combination of DCS and ERI models could improve the risk estimate of depressive symptoms in humans.展开更多
The deep deterministic policy gradient(DDPG)algo-rithm is an off-policy method that combines two mainstream reinforcement learning methods based on value iteration and policy iteration.Using the DDPG algorithm,agents ...The deep deterministic policy gradient(DDPG)algo-rithm is an off-policy method that combines two mainstream reinforcement learning methods based on value iteration and policy iteration.Using the DDPG algorithm,agents can explore and summarize the environment to achieve autonomous deci-sions in the continuous state space and action space.In this paper,a cooperative defense with DDPG via swarms of unmanned aerial vehicle(UAV)is developed and validated,which has shown promising practical value in the effect of defending.We solve the sparse rewards problem of reinforcement learning pair in a long-term task by building the reward function of UAV swarms and optimizing the learning process of artificial neural network based on the DDPG algorithm to reduce the vibration in the learning process.The experimental results show that the DDPG algorithm can guide the UAVs swarm to perform the defense task efficiently,meeting the requirements of a UAV swarm for non-centralization,autonomy,and promoting the intelligent development of UAVs swarm as well as the decision-making process.展开更多
Reward-based decision-making has been found to activate several brain areas, including the ven- trolateral prefronta~ lobe, orbitofrontal cortex, anterior cingulate cortex, ventral striatum, and mesolimbic dopaminergi...Reward-based decision-making has been found to activate several brain areas, including the ven- trolateral prefronta~ lobe, orbitofrontal cortex, anterior cingulate cortex, ventral striatum, and mesolimbic dopaminergic system. In this study, we observed brain areas activated under three de- grees of uncertainty in a reward-based decision-making task （certain, risky, and ambiguous）. The tasks were presented using a brain function audiovisual stimulation system. We conducted brain scans of 15 healthy volunteers using a 3.0T magnetic resonance scanner. We used SPM8 to ana- lyze the location and intensity of activation during the reward-based decision-making task, with re- spect to the three conditions. We found that the orbitofrontal cortex was activated in the certain reward condition, while the prefrontal cortex, precentral gyrus, occipital visual cortex, inferior parietal lobe, cerebellar posterior lobe, middle temporal gyrus, inferior temporal gyrus, limbic lobe, and midbrain were activated during the ＇risk＇ condition. The prefrontal cortex, temporal pole, inferior temporal gyrus, occipital visual cortex, and cerebellar posterior lobe were activated during am- biguous decision-making. The ventrolateral prefrontal lobe, frontal pole of the prefrontal lobe, orbi- tofrontal cortex, precentral gyrus, inferior temporal gyrus, fusiform gyrus, supramarginal gyrus, infe- rior parietal Iobule, and cerebellar posterior lobe exhibited greater activation in the ＇risk＇ than in the ＇certain＇ condition （P 〈 0.05）. The frontal pole and dorsolateral region of the prefrontal lobe, as well as the cerebellar posterior lobe, showed significantly greater activation in the ＇ambiguous＇ condition compared to the ＇risk＇ condition （P 〈 0.05）. The prefrontal lobe, occipital lobe, parietal lobe, temporal lobe, limbic lobe, midbrain, and posterior lobe of the cerebellum were activated during deci- sion-making about uncertain rewards. Thus, we observed different levels and regions of activation for different types of reward processing during decision-making. Specifically, when the degree of reward uncertainty increased, the number of activated brain areas increased, including greater ac- tivation of brain areas associated with loss.展开更多
The blades of wind turbines located at high latitudes are often covered with ice in late autumn and winter,where this affects their capacity for power generation as well as their safety.Accurately identifying the icin...The blades of wind turbines located at high latitudes are often covered with ice in late autumn and winter,where this affects their capacity for power generation as well as their safety.Accurately identifying the icing of the blades of wind turbines in remote areas is thus important,and a general model is needed to this end.This paper proposes a universal model based on a Deep Neural Network(DNN)that uses data from the Supervisory Control and Data Acquisition(SCADA)system.Two datasets from SCADA are first preprocessed through undersampling,that is,they are labeled,normalized,and balanced.The features of icing of the blades of a turbine identified in previous studies are then used to extract training data from the training dataset.A middle feature is proposed to show how a given feature is correlated with icing on the blade.Performance indicators for the model,including a reward function,are also designed to assess its predictive accuracy.Finally,the most suitable model is used to predict the testing data,and values of the reward function and the predictive accuracy of the model are calculated.The proposed method can be used to relate continuously transferred features with a binary status of icing of the blades of the turbine by using variables of the middle feature.The results here show that an integrated indicator systemis superior to a single indicator of accuracy when evaluating the prediction model.展开更多
In order to make strategic decision on firms’ sharing reward program( SRP), a nested Stackelberg game is developed. The sharing behavior among users and the rewarding strategy of firms are modeled. The optimal sharin...In order to make strategic decision on firms’ sharing reward program( SRP), a nested Stackelberg game is developed. The sharing behavior among users and the rewarding strategy of firms are modeled. The optimal sharing bonus is worked out and the impact of social relationships among customers is discussed. The results show that the higher the bonus,the more efforts the inductor is willing to make to persuade the inductee into buying. In addition,the firms should take the social relationship into consideration when setting the optimal sharing bonus. If the social relationship is weak,there is no need to adopt the SRP. Otherwise,there are two ways to reward the inductors. Also,the stronger the social relationship,the fewer the sharing bonuses that should be offered to the inductors,and the higher the expected profits. As a result,it is reasonable for the firms to implement SRPs on the social media where users are familiar with each other.展开更多
基金the National Natural Science Foun-dation of China(Grant No.71961003).
文摘In public goods games, punishments and rewards have been shown to be effective mechanisms for maintaining individualcooperation. However, punishments and rewards are costly to incentivize cooperation. Therefore, the generation ofcostly penalties and rewards has been a complex problem in promoting the development of cooperation. In real society,specialized institutions exist to punish evil people or reward good people by collecting taxes. We propose a strong altruisticpunishment or reward strategy in the public goods game through this phenomenon. Through theoretical analysis and numericalcalculation, we can get that tax-based strong altruistic punishment (reward) has more evolutionary advantages thantraditional strong altruistic punishment (reward) in maintaining cooperation and tax-based strong altruistic reward leads toa higher level of cooperation than tax-based strong altruistic punishment.
基金supported by the National Natural Science Foundation of China(71973001).
文摘To explore the green development of automobile enterprises and promote the achievement of the“dual carbon”target,based on the bounded rationality assumptions,this study constructed a tripartite evolutionary game model of gov-ernment,commercial banks,and automobile enterprises;introduced a dynamic reward and punishment mechanism;and analyzed the development process of the three parties’strategic behavior under the static and dynamic reward and punish-ment mechanism.Vensim PLE was used for numerical simulation analysis.Our results indicate that the system could not reach a stable state under the static reward and punishment mechanism.A dynamic reward and punishment mechanism can effectively improve the system stability and better fit real situations.Under the dynamic reward and punishment mechan-ism,an increase in the initial probabilities of the three parties can promote the system stability,and the government can im-plement effective supervision by adjusting the upper limit of the reward and punishment intensity.Finally,the implementa-tion of green credit by commercial banks plays a significant role in promoting the green development of automobile enter-prises.
基金funded by National Natural Science Foundation of China(No.62063006)Guangxi Science and Technology Major Program(No.2022AA05002)+1 种基金Key Laboratory of AI and Information Processing(Hechi University),Education Department of Guangxi Zhuang Autonomous Region(No.2022GXZDSY003)Central Leading Local Science and Technology Development Fund Project of Wuzhou(No.202201001).
文摘By integrating deep neural networks with reinforcement learning,the Double Deep Q Network(DDQN)algorithm overcomes the limitations of Q-learning in handling continuous spaces and is widely applied in the path planning of mobile robots.However,the traditional DDQN algorithm suffers from sparse rewards and inefficient utilization of high-quality data.Targeting those problems,an improved DDQN algorithm based on average Q-value estimation and reward redistribution was proposed.First,to enhance the precision of the target Q-value,the average of multiple previously learned Q-values from the target Q network is used to replace the single Q-value from the current target Q network.Next,a reward redistribution mechanism is designed to overcome the sparse reward problem by adjusting the final reward of each action using the round reward from trajectory information.Additionally,a reward-prioritized experience selection method is introduced,which ranks experience samples according to reward values to ensure frequent utilization of high-quality data.Finally,simulation experiments are conducted to verify the effectiveness of the proposed algorithm in fixed-position scenario and random environments.The experimental results show that compared to the traditional DDQN algorithm,the proposed algorithm achieves shorter average running time,higher average return and fewer average steps.The performance of the proposed algorithm is improved by 11.43%in the fixed scenario and 8.33%in random environments.It not only plans economic and safe paths but also significantly improves efficiency and generalization in path planning,making it suitable for widespread application in autonomous navigation and industrial automation.
文摘Cross-lingual image description,the task of generating image captions in a target language from images and descriptions in a source language,is addressed in this study through a novel approach that combines neural network models and semantic matching techniques.Experiments conducted on the Flickr8k and AraImg2k benchmark datasets,featuring images and descriptions in English and Arabic,showcase remarkable performance improvements over state-of-the-art methods.Our model,equipped with the Image&Cross-Language Semantic Matching module and the Target Language Domain Evaluation module,significantly enhances the semantic relevance of generated image descriptions.For English-to-Arabic and Arabic-to-English cross-language image descriptions,our approach achieves a CIDEr score for English and Arabic of 87.9%and 81.7%,respectively,emphasizing the substantial contributions of our methodology.Comparative analyses with previous works further affirm the superior performance of our approach,and visual results underscore that our model generates image captions that are both semantically accurate and stylistically consistent with the target language.In summary,this study advances the field of cross-lingual image description,offering an effective solution for generating image captions across languages,with the potential to impact multilingual communication and accessibility.Future research directions include expanding to more languages and incorporating diverse visual and textual data sources.
基金supported by the Key Research and Development Program of Shaanxi (2022GXLH-02-09)the Aeronautical Science Foundation of China (20200051053001)the Natural Science Basic Research Program of Shaanxi (2020JM-147)。
文摘Autonomous umanned aerial vehicle(UAV) manipulation is necessary for the defense department to execute tactical missions given by commanders in the future unmanned battlefield. A large amount of research has been devoted to improving the autonomous decision-making ability of UAV in an interactive environment, where finding the optimal maneuvering decisionmaking policy became one of the key issues for enabling the intelligence of UAV. In this paper, we propose a maneuvering decision-making algorithm for autonomous air-delivery based on deep reinforcement learning under the guidance of expert experience. Specifically, we refine the guidance towards area and guidance towards specific point tasks for the air-delivery process based on the traditional air-to-surface fire control methods.Moreover, we construct the UAV maneuvering decision-making model based on Markov decision processes(MDPs). Specifically, we present a reward shaping method for the guidance towards area and guidance towards specific point tasks using potential-based function and expert-guided advice. The proposed algorithm could accelerate the convergence of the maneuvering decision-making policy and increase the stability of the policy in terms of the output during the later stage of training process. The effectiveness of the proposed maneuvering decision-making policy is illustrated by the curves of training parameters and extensive experimental results for testing the trained policy.
文摘Mobile adhoc networks have grown in prominence in recent years,and they are now utilized in a broader range of applications.The main challenges are related to routing techniques that are generally employed in them.Mobile Adhoc system management,on the other hand,requires further testing and improvements in terms of security.Traditional routing protocols,such as Adhoc On-Demand Distance Vector(AODV)and Dynamic Source Routing(DSR),employ the hop count to calculate the distance between two nodes.The main aim of this research work is to determine the optimum method for sending packets while also extending life time of the network.It is achieved by changing the residual energy of each network node.Also,in this paper,various algorithms for optimal routing based on parameters like energy,distance,mobility,and the pheromone value are proposed.Moreover,an approach based on a reward and penalty system is given in this paper to evaluate the efficiency of the proposed algorithms under the impact of parameters.The simulation results unveil that the reward penalty-based approach is quite effective for the selection of an optimal path for routing when the algorithms are implemented under the parameters of interest,which helps in achieving less packet drop and energy consumption of the nodes along with enhancing the network efficiency.
基金supported in part by the National Natural Science Foundation of China(62006111,62073160)the Natural Science Foundation of Jiangsu Province of China(BK20200330)。
文摘Goal-conditioned reinforcement learning(RL)is an interesting extension of the traditional RL framework,where the dynamic environment and reward sparsity can cause conventional learning algorithms to fail.Reward shaping is a practical approach to improving sample efficiency by embedding human domain knowledge into the learning process.Existing reward shaping methods for goal-conditioned RL are typically built on distance metrics with a linear and isotropic distribution,which may fail to provide sufficient information about the ever-changing environment with high complexity.This paper proposes a novel magnetic field-based reward shaping(MFRS)method for goal-conditioned RL tasks with dynamic target and obstacles.Inspired by the physical properties of magnets,we consider the target and obstacles as permanent magnets and establish the reward function according to the intensity values of the magnetic field generated by these magnets.The nonlinear and anisotropic distribution of the magnetic field intensity can provide more accessible and conducive information about the optimization landscape,thus introducing a more sophisticated magnetic reward compared to the distance-based setting.Further,we transform our magnetic reward to the form of potential-based reward shaping by learning a secondary potential function concurrently to ensure the optimal policy invariance of our method.Experiments results in both simulated and real-world robotic manipulation tasks demonstrate that MFRS outperforms relevant existing methods and effectively improves the sample efficiency of RL algorithms in goal-conditioned tasks with various dynamics of the target and obstacles.
文摘As assessment outcomes provide students with a sense of accomplishment that is boosted by the reward system,learning becomes more effective.This research aims to determine the effects of reward system prior to assessment in Mathematics.Quasi-experimental research design was used to examine whether there was a significant difference between the use of reward system and students’level of performance in Mathematics.Through purposive sampling,the respondents of the study involve 80 Grade 9 students belonging to two sections from Gaudencio B.Lontok Memorial Integrated School.Based on similar demographics and pre-test results,control and study group were involved as participants of the study.Data were treated and analyzed accordingly using statistical treatments such as mean and t-test for independent variables.There was a significant finding revealing the advantage of using the reward system compare to the non-reward system in increasing students’level of performance in Mathematics.It is concluded that the use of reward system is effective in improving the assessment outcomes in Mathematics.It is recommended to use reward system for persistent assessment outcomes prior to assessment,to be a reflection of the intended outcomes in Mathematics.
基金Supported by National Natural Science Funds of China(71503251)The Agricultural Science and Technology Innovation Program(ASTIP-IAED-2015-01)China forage and grass research system(CARS-35-22)~~
文摘Grassland ecological protection compensation and reward policy is the largest-scale investment concerning themost extensive areas since foundation of the PRC. It will be the long-term implementationpolicy for grassland ecological protection. In this study,based on macro-perspective, the policy effects ofgrasslandproductivity, ecological protection, animal husbandryoutput, pastoralists＇ income were ana- lyzed. The resultsshow that, afterimplementation of the policy, naturalgrass production and grasslandtheoretical stocking rateincreased. The averagenatural grasslandlivestockoverloading ratedecreased significantly, comprehensivenationalgrasslandvegetation coverageis increasing. Besides, adult cattleandbeef yield arefluctuated. Sheep head, adult sheep, sheep production, milk productionincreasedin varying degrees. The per capita netincomeof farmers and pastoralists, livestock income, the proportion oflivestockincomewere higher than those beforeimplementation of the policy.
基金National Science Foundation of China (3047055330530270+10 种基金30670669 30770700)973 Program (2005CB522803 2007CB947703)863 Program (O7013810 2006AA02A116)The Major State Basic Research of China (2003CB716600)Chinese-Finnish International Collaboration Project-neuro (30621130076)Program of CASC (KSCX1-YW-R-33YZ200737)National Key Technologies R & D Program and Yunnan Science and Technique Program (2006PT08-2)
文摘The orbitofrontal cortex （OFC） is particularly important for the neural representation of reward value. Previous studies indicated that electroencephalogram （EEG） activity in the OFC was involved in drug administration and withdrawal. The present study investigated EEG activity in the OFC in rats during the development of food reward and craving. Two environments were used separately for control and food-related EEG recordings. In the food-related environment rats were first trained to eat chocolate peanuts; then they either had no access to this food, but could see and smell it （craving trials）, or had free access to this food （reward trials）. The EEG in the left OFC was recorded during these trials. We showed that, in the food-related environment the EEG activity peaking in the delta band （2-4 Hz） was significantly correlated with the stimulus, increasing during food reward and decreasing during food craving when compared with that in the control environment. Our data suggests that EEG activity in the OFC can be altered by food reward; moreover, delta rhythm in this region could be used as an index monitoring changed signal underlying this reward.
基金funded by Henan Provincial Health Science and Technology Key Projects(201001009)National Science and Technology Infrastructure Program(2006BAI06B 08),China
文摘Objective To investigate the co-effect of Demand-control-support （DCS） model and Effort-reward Imbalance （ERI） model on the risk estimation of depression in humans in comparison with the effects when they are used respectively. Methods A total of 3 632 males and 1 706 females from 13 factories and companies in Henan province were recruited in this cross-sectional study. Perceived job stress was evaluated with the Job Content Questionnaire and Effort-Reward Imbalance Questionnaire （Chinese version）. Depressive symptoms were assessed by using the Center for Epidemiological Studies Depression Scale （CES-D）. Results DC （demands/job control ratio） and ERI were shown to be independently associated with depressive symptoms. The outcome of low social support and overcommitment were similar. High DC and low social support （SS）, high ERI and high overcommitment, and high DC and high ERI posed greater risks of depressive symptoms than each of them did alone. ERI model and SS model seem to be effective in estimating the risk of depressive symptoms if they are used respectively. Conclusion The DC had better performance when it was used in combination with low SS. The effect on physical demands was better than on psychological demands. The combination of DCS and ERI models could improve the risk estimate of depressive symptoms in humans.
基金supported by the Key Research and Development Program of Shaanxi(2022GY-089)the Natural Science Basic Research Program of Shaanxi(2022JQ-593).
文摘The deep deterministic policy gradient(DDPG)algo-rithm is an off-policy method that combines two mainstream reinforcement learning methods based on value iteration and policy iteration.Using the DDPG algorithm,agents can explore and summarize the environment to achieve autonomous deci-sions in the continuous state space and action space.In this paper,a cooperative defense with DDPG via swarms of unmanned aerial vehicle(UAV)is developed and validated,which has shown promising practical value in the effect of defending.We solve the sparse rewards problem of reinforcement learning pair in a long-term task by building the reward function of UAV swarms and optimizing the learning process of artificial neural network based on the DDPG algorithm to reduce the vibration in the learning process.The experimental results show that the DDPG algorithm can guide the UAVs swarm to perform the defense task efficiently,meeting the requirements of a UAV swarm for non-centralization,autonomy,and promoting the intelligent development of UAVs swarm as well as the decision-making process.
基金supported by the Science and Technology Development Project of Shandong Province,China,No.2011YD18045the Natural Science Foundation of Shandong Province,China,No.ZR2012HM049+3 种基金the Health Care Foundation Program of Shandong Province,China,No.2007BZ19the Foundation Program of Technology Bureau of Qingdao,ChinaNo.Kzd-0309-1-1-33-nsh
文摘Reward-based decision-making has been found to activate several brain areas, including the ven- trolateral prefronta~ lobe, orbitofrontal cortex, anterior cingulate cortex, ventral striatum, and mesolimbic dopaminergic system. In this study, we observed brain areas activated under three de- grees of uncertainty in a reward-based decision-making task （certain, risky, and ambiguous）. The tasks were presented using a brain function audiovisual stimulation system. We conducted brain scans of 15 healthy volunteers using a 3.0T magnetic resonance scanner. We used SPM8 to ana- lyze the location and intensity of activation during the reward-based decision-making task, with re- spect to the three conditions. We found that the orbitofrontal cortex was activated in the certain reward condition, while the prefrontal cortex, precentral gyrus, occipital visual cortex, inferior parietal lobe, cerebellar posterior lobe, middle temporal gyrus, inferior temporal gyrus, limbic lobe, and midbrain were activated during the ＇risk＇ condition. The prefrontal cortex, temporal pole, inferior temporal gyrus, occipital visual cortex, and cerebellar posterior lobe were activated during am- biguous decision-making. The ventrolateral prefrontal lobe, frontal pole of the prefrontal lobe, orbi- tofrontal cortex, precentral gyrus, inferior temporal gyrus, fusiform gyrus, supramarginal gyrus, infe- rior parietal Iobule, and cerebellar posterior lobe exhibited greater activation in the ＇risk＇ than in the ＇certain＇ condition （P 〈 0.05）. The frontal pole and dorsolateral region of the prefrontal lobe, as well as the cerebellar posterior lobe, showed significantly greater activation in the ＇ambiguous＇ condition compared to the ＇risk＇ condition （P 〈 0.05）. The prefrontal lobe, occipital lobe, parietal lobe, temporal lobe, limbic lobe, midbrain, and posterior lobe of the cerebellum were activated during deci- sion-making about uncertain rewards. Thus, we observed different levels and regions of activation for different types of reward processing during decision-making. Specifically, when the degree of reward uncertainty increased, the number of activated brain areas increased, including greater ac- tivation of brain areas associated with loss.
基金supported by the National Natural Science Foundation of China under Grant No.61573138.
文摘The blades of wind turbines located at high latitudes are often covered with ice in late autumn and winter,where this affects their capacity for power generation as well as their safety.Accurately identifying the icing of the blades of wind turbines in remote areas is thus important,and a general model is needed to this end.This paper proposes a universal model based on a Deep Neural Network(DNN)that uses data from the Supervisory Control and Data Acquisition(SCADA)system.Two datasets from SCADA are first preprocessed through undersampling,that is,they are labeled,normalized,and balanced.The features of icing of the blades of a turbine identified in previous studies are then used to extract training data from the training dataset.A middle feature is proposed to show how a given feature is correlated with icing on the blade.Performance indicators for the model,including a reward function,are also designed to assess its predictive accuracy.Finally,the most suitable model is used to predict the testing data,and values of the reward function and the predictive accuracy of the model are calculated.The proposed method can be used to relate continuously transferred features with a binary status of icing of the blades of the turbine by using variables of the middle feature.The results here show that an integrated indicator systemis superior to a single indicator of accuracy when evaluating the prediction model.
基金The National Social Science Foundation of China(No.17BGL196)the Postgraduate Research&Practice Innovation Program of Jiangsu Province(No.KYLX15_0193)
文摘In order to make strategic decision on firms’ sharing reward program( SRP), a nested Stackelberg game is developed. The sharing behavior among users and the rewarding strategy of firms are modeled. The optimal sharing bonus is worked out and the impact of social relationships among customers is discussed. The results show that the higher the bonus,the more efforts the inductor is willing to make to persuade the inductee into buying. In addition,the firms should take the social relationship into consideration when setting the optimal sharing bonus. If the social relationship is weak,there is no need to adopt the SRP. Otherwise,there are two ways to reward the inductors. Also,the stronger the social relationship,the fewer the sharing bonuses that should be offered to the inductors,and the higher the expected profits. As a result,it is reasonable for the firms to implement SRPs on the social media where users are familiar with each other.
