Reinforcement Learning(RL)techniques are being studied to solve the Demand and Capacity Balancing(DCB)problems to fully exploit their computational performance.A locally gen-eralised Multi-Agent Reinforcement Learning...Reinforcement Learning(RL)techniques are being studied to solve the Demand and Capacity Balancing(DCB)problems to fully exploit their computational performance.A locally gen-eralised Multi-Agent Reinforcement Learning(MARL)for real-world DCB problems is proposed.The proposed method can deploy trained agents directly to unseen scenarios in a specific Air Traffic Flow Management(ATFM)region to quickly obtain a satisfactory solution.In this method,agents of all flights in a scenario form a multi-agent decision-making system based on partial observation.The trained agent with the customised neural network can be deployed directly on the corresponding flight,allowing it to solve the DCB problem jointly.A cooperation coefficient is introduced in the reward function,which is used to adjust the agent’s cooperation preference in a multi-agent system,thereby controlling the distribution of flight delay time allocation.A multi-iteration mechanism is designed for the DCB decision-making framework to deal with problems arising from non-stationarity in MARL and to ensure that all hotspots are eliminated.Experiments based on large-scale high-complexity real-world scenarios are conducted to verify the effectiveness and efficiency of the method.From a statis-tical point of view,it is proven that the proposed method is generalised within the scope of the flights and sectors of interest,and its optimisation performance outperforms the standard computer-assisted slot allocation and state-of-the-art RL-based DCB methods.The sensitivity analysis preliminarily reveals the effect of the cooperation coefficient on delay time allocation.展开更多
Models of adaptive behaviour typically assume that animals behave as though they have highly complex, detailed strategies for making decisions. In reality, selection favours the optimal balance between the costs and b...Models of adaptive behaviour typically assume that animals behave as though they have highly complex, detailed strategies for making decisions. In reality, selection favours the optimal balance between the costs and benefits of complexity. Here we investigate this trade-off for an animal that has to decide whether or not to forage for food - and so how much energy reserves to store - depending on the food availability in its environment. We evolve a decision rule that controls the target reserve level for different ranges of food availability, but where increasing complexity is costly in that metabolic rate increases with the sensitivity of the rule. The evolved rule tends to be much less complex than the optimal strategy but performs almost as well, while being less costly to implement. It achieves this by being highly sensitive to changing food availability at low food abun- dance - where it provides a close fit to the optimal strategy - but insensitive when food is plentiful. When food availability is high, the target reserve level that evolves is much higher than under the optimal strategy, which has implications for our under- standing of obesity. Our work highlights the important principle of generalisability of simple decision-making mechanisms, which enables animals to respond reasonably well to conditions not directly experienced by themselves or their ancestors.展开更多
Background:It is important to determine prognostic factors for the outcome of amyotrophic lateral sclerosis (ALS) at an early stage.The time taken for symptoms to spread from spinal or bulbar regions to both (time to ...Background:It is important to determine prognostic factors for the outcome of amyotrophic lateral sclerosis (ALS) at an early stage.The time taken for symptoms to spread from spinal or bulbar regions to both (time to generalization;TTG) is considered a strong predictor of survival;however,this has rarely been studied in Asian populations.The aim of this retrospective study was to evaluate potential factors affecting prognosis in Chinese patients with sporadic ALS,with a focus on the association between TTG and overall survival.Methods:Seventy-one patients with sporadic ALS who were hospitalized at Chinese PLA General Hospital from 2009 to 2016 were followed up until December 2017.Survival analysis was performed using univariate Kaplan-Meier log-rank and multivariate Cox proportional hazards models.The clinical data of the patients were recorded and analyzed.Variables studied were age at symptom onset,sex,site of symptom onset,diagnostic latency,TTG,diagnostic category,ALS Functional Rating Scale-revised score,percent predicted forced vital capacity (FVC%),and disease progression rate (DPR) at diagnosis.Results:The mean age at onset was 54 (SD = 10.2) years,and the median survival time from symptom onset was 41 months (95% confidence interval:34–47).By univariate analysis,factors independently affecting survival were age at symptom onset (Log rank = 15.652,P<0.0001),TTG (Log rank = 14.728,P<0.0001),diagnostic latency (Log rank = 11.997,P = 0.001),and DPR (Log rank = 6.50,P = 0.011).In the Cox multivariate model,TTG had the strongest impact on survival time (hazard ratio = 0.926,P = 0.01).Conclusions:TTG can be used as an effective indicator of prognosis in patients with sporadic ALS.展开更多
This paper presents an overview of research studies made at the COGIT laboratory of IGN France in the fields of generalisation and symbol specification,particularly considering evaluation aspects.It then discusses how...This paper presents an overview of research studies made at the COGIT laboratory of IGN France in the fields of generalisation and symbol specification,particularly considering evaluation aspects.It then discusses how generalisation and symbol specification interact.Finally it explores some possible adaptations of the presented works in generalisation and symbol specification to cartography in the context of crisis management.展开更多
This paper presents several neural network based modelling, reliable optimal control, and iterative learning control methods for batch processes. In order to overcome the lack of robustness of a single neural network,...This paper presents several neural network based modelling, reliable optimal control, and iterative learning control methods for batch processes. In order to overcome the lack of robustness of a single neural network, bootstrap aggregated neural networks are used to build reliable data based empirical models. Apart from improving the model generalisation capability, a bootstrap aggregated neural network can also provide model prediction confidence bounds. A reliable optimal control method by incorporating model prediction confidence bounds into the optimisation objective function is presented. A neural network based iterative learning control strategy is presented to overcome the problem due to unknown disturbances and model-plant mismatches. The proposed methods are demonstrated on a simulated batch polymerisation process.展开更多
基金co-funded by the National Natural Science Foundation of China(No.61903187)the National Key R&D Program of China(No.2021YFB1600500)+2 种基金the China Scholarship Council(No.202006830095)the Natural Science Foundation of Jiangsu Province(No.BK20190414)the Jiangsu Province Postgraduate Innovation Fund(No.KYCX20_0213).
文摘Reinforcement Learning(RL)techniques are being studied to solve the Demand and Capacity Balancing(DCB)problems to fully exploit their computational performance.A locally gen-eralised Multi-Agent Reinforcement Learning(MARL)for real-world DCB problems is proposed.The proposed method can deploy trained agents directly to unseen scenarios in a specific Air Traffic Flow Management(ATFM)region to quickly obtain a satisfactory solution.In this method,agents of all flights in a scenario form a multi-agent decision-making system based on partial observation.The trained agent with the customised neural network can be deployed directly on the corresponding flight,allowing it to solve the DCB problem jointly.A cooperation coefficient is introduced in the reward function,which is used to adjust the agent’s cooperation preference in a multi-agent system,thereby controlling the distribution of flight delay time allocation.A multi-iteration mechanism is designed for the DCB decision-making framework to deal with problems arising from non-stationarity in MARL and to ensure that all hotspots are eliminated.Experiments based on large-scale high-complexity real-world scenarios are conducted to verify the effectiveness and efficiency of the method.From a statis-tical point of view,it is proven that the proposed method is generalised within the scope of the flights and sectors of interest,and its optimisation performance outperforms the standard computer-assisted slot allocation and state-of-the-art RL-based DCB methods.The sensitivity analysis preliminarily reveals the effect of the cooperation coefficient on delay time allocation.
文摘Models of adaptive behaviour typically assume that animals behave as though they have highly complex, detailed strategies for making decisions. In reality, selection favours the optimal balance between the costs and benefits of complexity. Here we investigate this trade-off for an animal that has to decide whether or not to forage for food - and so how much energy reserves to store - depending on the food availability in its environment. We evolve a decision rule that controls the target reserve level for different ranges of food availability, but where increasing complexity is costly in that metabolic rate increases with the sensitivity of the rule. The evolved rule tends to be much less complex than the optimal strategy but performs almost as well, while being less costly to implement. It achieves this by being highly sensitive to changing food availability at low food abun- dance - where it provides a close fit to the optimal strategy - but insensitive when food is plentiful. When food availability is high, the target reserve level that evolves is much higher than under the optimal strategy, which has implications for our under- standing of obesity. Our work highlights the important principle of generalisability of simple decision-making mechanisms, which enables animals to respond reasonably well to conditions not directly experienced by themselves or their ancestors.
基金grants from the National Science Foundation of China (No.81671278)"One Hundred Advantage Projects" Fund of Chinese PLA General Hospital (No.YS201415).
文摘Background:It is important to determine prognostic factors for the outcome of amyotrophic lateral sclerosis (ALS) at an early stage.The time taken for symptoms to spread from spinal or bulbar regions to both (time to generalization;TTG) is considered a strong predictor of survival;however,this has rarely been studied in Asian populations.The aim of this retrospective study was to evaluate potential factors affecting prognosis in Chinese patients with sporadic ALS,with a focus on the association between TTG and overall survival.Methods:Seventy-one patients with sporadic ALS who were hospitalized at Chinese PLA General Hospital from 2009 to 2016 were followed up until December 2017.Survival analysis was performed using univariate Kaplan-Meier log-rank and multivariate Cox proportional hazards models.The clinical data of the patients were recorded and analyzed.Variables studied were age at symptom onset,sex,site of symptom onset,diagnostic latency,TTG,diagnostic category,ALS Functional Rating Scale-revised score,percent predicted forced vital capacity (FVC%),and disease progression rate (DPR) at diagnosis.Results:The mean age at onset was 54 (SD = 10.2) years,and the median survival time from symptom onset was 41 months (95% confidence interval:34–47).By univariate analysis,factors independently affecting survival were age at symptom onset (Log rank = 15.652,P<0.0001),TTG (Log rank = 14.728,P<0.0001),diagnostic latency (Log rank = 11.997,P = 0.001),and DPR (Log rank = 6.50,P = 0.011).In the Cox multivariate model,TTG had the strongest impact on survival time (hazard ratio = 0.926,P = 0.01).Conclusions:TTG can be used as an effective indicator of prognosis in patients with sporadic ALS.
文摘This paper presents an overview of research studies made at the COGIT laboratory of IGN France in the fields of generalisation and symbol specification,particularly considering evaluation aspects.It then discusses how generalisation and symbol specification interact.Finally it explores some possible adaptations of the presented works in generalisation and symbol specification to cartography in the context of crisis management.
基金Supported by UK EPSRC (grants GR/N13319 and GR/R 10875)
文摘This paper presents several neural network based modelling, reliable optimal control, and iterative learning control methods for batch processes. In order to overcome the lack of robustness of a single neural network, bootstrap aggregated neural networks are used to build reliable data based empirical models. Apart from improving the model generalisation capability, a bootstrap aggregated neural network can also provide model prediction confidence bounds. A reliable optimal control method by incorporating model prediction confidence bounds into the optimisation objective function is presented. A neural network based iterative learning control strategy is presented to overcome the problem due to unknown disturbances and model-plant mismatches. The proposed methods are demonstrated on a simulated batch polymerisation process.
