The adopters of IoT face challenges with the surging Internet-based attacks on their IoT assets and inefficiencies within the technology. Unfortunately, IoT is overly distributed, still evolving and facing implementat...The adopters of IoT face challenges with the surging Internet-based attacks on their IoT assets and inefficiencies within the technology. Unfortunately, IoT is overly distributed, still evolving and facing implementation and security challenges. Given the above scenario, we argue that the IoT network should always be decentralized design, and security should be built by design. The paper is the design and construction of a decentralized IoT security framework, with the goal of making emerging IoT systems more resilient to attacks and supporting complex communication and resource sharing. The framework improves efficiency and scalability in IoT, exposes vulnerable subsystems and components as possible weak links to system compromise, and meets the requirements of a heterogeneous computing environment. Other features of the framework including efficient resource sharing, fault tolerance, and distributed storage support the Internet of Things. We discuss the design requirements and carry out the implementation of Proof of Concept and evaluation of our framework. Two underlying technologies: the actor model and the blockchain were used for the implementation. Our reason for choosing the actor model and blockchain is to compare its suitability for IoT integration in parallel. Hence, evaluation of the system is performed based on computational and memory efficiency, security, and scalability. We conclude from the evaluations that the actor-based implementation has better scalability than the block-chain-based implementation. Also, the blockchain seems to be computationally more intensive than the actors and less suitable for IoT systems.展开更多
Bamboo is an important non-timber forest product owing to its multipurpose nature. In Cameroon, bamboo has always been neglected and seen as worthless by many communities. However, in recent years</span></spa...Bamboo is an important non-timber forest product owing to its multipurpose nature. In Cameroon, bamboo has always been neglected and seen as worthless by many communities. However, in recent years</span></span><span><span><span style="font-family:"">,</span></span></span><span><span><span style="font-family:""> bamboo has received renewed attention which has made many communities and different stakeholders to gain interest in the resource. This study which was carried out in the Menoua division, West region of Cameroon between February to July 2021 sought to identify and characterise the actors involved in the bamboo sector, and assess the different strategies that could be employed to develop the bamboo sector. Data w</span></span></span><span><span><span style="font-family:"">ere</span></span></span><span><span><span style="font-family:""> collected using quantitative and qualitative methods in 6 sub-divisions in Menoua division (Dschang, Santchou, Fongo Tongo, Fokoué, Penka Michel and Nkong-Ni). For the selection of study sites, the criterion used was the proximity of households to the area where the bamboo resource is found. Using the simple random sampling method, 53 households were surveyed across the Menoua division. Key informant and expert interviews were also conducted with mayors, traditional chiefs, heads of forestry posts, heads of research institutions/structures and craftsmen involved in the bamboo sector. Using SPHINX software version 5, a survey form was designed. The data collected was coded and analysed using EXCEL 2010 and SPSS. Map data was analysed using ARCGIS version 2.18. Results indicated that there were two main groups of actors: direct actors (collectors, <span>collectors/transporters, producers/collectors, producers/collectors/transporters,</span> collectors/traders) and indirect actors (Municipal Councils, Decentralized Services of the Ministry of Forestry and Wildlife</span></span></span><span><span><span style="font-family:"">—<span>MINFOF, Development partners like INBAR, Research Institutions like the Institute of Agricultural Research for Development</span>—<span>IRAD, and academic institutions like the University of Dschang). The direct actors are directly linked to the bamboo value chain while the indirect actors are those whose decisions influence the sector (e.g. MINFOF) or those who provide financial and technical support (Municipal Councils, INBAR;IRAD, the University of Dschang). The main strategies proposed by the local population for the development of the bamboo sector were regular sensitization campaigns on the benefits of bamboo (92%) and the provision of technical, material and financial support to bamboo producers (41%). Key informants/experts proposed the following in order to ensure the development of the bamboo sector in the Menoua division: raising awareness about bamboo, its different varieties and benefits;creating bamboo plantations (with varieties adapted to the agro-ecological zone i.e. the western highlands) in order to reduce the pressure on other resources;setting up support mechanisms for producers and other actors in the bamboo value chain;allocating land/agricultural areas for bamboo plantations as there is land scarcity in the Menoua division. Based on the strategic framework developed from this study, in order to ensure an adequate and effective development of the bamboo sector in the Menoua division, there should be among others: multiplication of awareness-raising and training programmes for farmers on bamboo production techniques;more support for smallholder farmers by providing them bamboo plants in quality and quantity;production of bamboo stems in quality and quantity;more awareness campaigns for young craftsmen on the advantages of the bamboo craft sector;more training campaigns for craftsmen on modern bamboo processing techniques;and the establishment of a well-developed and sustainable bamboo-based craft sector.展开更多
离线强化学习(Offline RL)定义了从固定批次的数据集中学习的任务,能够规避与环境交互的风险,提高学习的效率与稳定性。其中优势加权行动者-评论家算法提出了一种将样本高效动态规划与最大似然策略更新相结合的方法,在利用大量离线数据...离线强化学习(Offline RL)定义了从固定批次的数据集中学习的任务,能够规避与环境交互的风险,提高学习的效率与稳定性。其中优势加权行动者-评论家算法提出了一种将样本高效动态规划与最大似然策略更新相结合的方法,在利用大量离线数据的同时,快速执行在线精细化策略的调整。但是该算法使用随机经验回放机制,同时行动者-评论家模型只采用一套行动者,数据采样与回放不平衡。针对以上问题,提出一种基于策略蒸馏并进行数据经验优选回放的优势加权双行动者-评论家算法(Advantage Weighted Double Actors-Critics Based on Policy Distillation with Data Experience Optimization and Replay,DOR-PDAWAC),该算法采用偏好新经验并重复回放新旧经验的机制,利用双行动者增加探索,并运用基于策略蒸馏的主从框架,将行动者分为主行为者和从行为者,提升协作效率。将所提算法应用到通用D4RL数据集中的MuJoCo任务上进行消融实验与对比实验,结果表明,其学习效率等均获得了更优的表现。展开更多
深度强化学习在训练过程中会探索大量环境样本,造成算法收敛时间过长,而重用或传输来自先前任务(源任务)学习的知识,对算法在新任务(目标任务)的学习具有提高算法收敛速度的潜力。为了提高算法学习效率,提出一种双Q网络学习的迁移强化...深度强化学习在训练过程中会探索大量环境样本,造成算法收敛时间过长,而重用或传输来自先前任务(源任务)学习的知识,对算法在新任务(目标任务)的学习具有提高算法收敛速度的潜力。为了提高算法学习效率,提出一种双Q网络学习的迁移强化学习算法,其基于actor-critic框架迁移源任务最优值函数的知识,使目标任务中值函数网络对策略作出更准确的评价,引导策略快速向最优策略方向更新。将该算法用于Open AI Gym以及在三维空间机械臂到达目标物位置的实验中,相比于常规深度强化学习算法取得了更好的效果,实验证明提出的双Q网络学习的迁移强化学习算法具有较快的收敛速度,并且在训练过程中算法探索更加稳定。展开更多
文摘The adopters of IoT face challenges with the surging Internet-based attacks on their IoT assets and inefficiencies within the technology. Unfortunately, IoT is overly distributed, still evolving and facing implementation and security challenges. Given the above scenario, we argue that the IoT network should always be decentralized design, and security should be built by design. The paper is the design and construction of a decentralized IoT security framework, with the goal of making emerging IoT systems more resilient to attacks and supporting complex communication and resource sharing. The framework improves efficiency and scalability in IoT, exposes vulnerable subsystems and components as possible weak links to system compromise, and meets the requirements of a heterogeneous computing environment. Other features of the framework including efficient resource sharing, fault tolerance, and distributed storage support the Internet of Things. We discuss the design requirements and carry out the implementation of Proof of Concept and evaluation of our framework. Two underlying technologies: the actor model and the blockchain were used for the implementation. Our reason for choosing the actor model and blockchain is to compare its suitability for IoT integration in parallel. Hence, evaluation of the system is performed based on computational and memory efficiency, security, and scalability. We conclude from the evaluations that the actor-based implementation has better scalability than the block-chain-based implementation. Also, the blockchain seems to be computationally more intensive than the actors and less suitable for IoT systems.
文摘Bamboo is an important non-timber forest product owing to its multipurpose nature. In Cameroon, bamboo has always been neglected and seen as worthless by many communities. However, in recent years</span></span><span><span><span style="font-family:"">,</span></span></span><span><span><span style="font-family:""> bamboo has received renewed attention which has made many communities and different stakeholders to gain interest in the resource. This study which was carried out in the Menoua division, West region of Cameroon between February to July 2021 sought to identify and characterise the actors involved in the bamboo sector, and assess the different strategies that could be employed to develop the bamboo sector. Data w</span></span></span><span><span><span style="font-family:"">ere</span></span></span><span><span><span style="font-family:""> collected using quantitative and qualitative methods in 6 sub-divisions in Menoua division (Dschang, Santchou, Fongo Tongo, Fokoué, Penka Michel and Nkong-Ni). For the selection of study sites, the criterion used was the proximity of households to the area where the bamboo resource is found. Using the simple random sampling method, 53 households were surveyed across the Menoua division. Key informant and expert interviews were also conducted with mayors, traditional chiefs, heads of forestry posts, heads of research institutions/structures and craftsmen involved in the bamboo sector. Using SPHINX software version 5, a survey form was designed. The data collected was coded and analysed using EXCEL 2010 and SPSS. Map data was analysed using ARCGIS version 2.18. Results indicated that there were two main groups of actors: direct actors (collectors, <span>collectors/transporters, producers/collectors, producers/collectors/transporters,</span> collectors/traders) and indirect actors (Municipal Councils, Decentralized Services of the Ministry of Forestry and Wildlife</span></span></span><span><span><span style="font-family:"">—<span>MINFOF, Development partners like INBAR, Research Institutions like the Institute of Agricultural Research for Development</span>—<span>IRAD, and academic institutions like the University of Dschang). The direct actors are directly linked to the bamboo value chain while the indirect actors are those whose decisions influence the sector (e.g. MINFOF) or those who provide financial and technical support (Municipal Councils, INBAR;IRAD, the University of Dschang). The main strategies proposed by the local population for the development of the bamboo sector were regular sensitization campaigns on the benefits of bamboo (92%) and the provision of technical, material and financial support to bamboo producers (41%). Key informants/experts proposed the following in order to ensure the development of the bamboo sector in the Menoua division: raising awareness about bamboo, its different varieties and benefits;creating bamboo plantations (with varieties adapted to the agro-ecological zone i.e. the western highlands) in order to reduce the pressure on other resources;setting up support mechanisms for producers and other actors in the bamboo value chain;allocating land/agricultural areas for bamboo plantations as there is land scarcity in the Menoua division. Based on the strategic framework developed from this study, in order to ensure an adequate and effective development of the bamboo sector in the Menoua division, there should be among others: multiplication of awareness-raising and training programmes for farmers on bamboo production techniques;more support for smallholder farmers by providing them bamboo plants in quality and quantity;production of bamboo stems in quality and quantity;more awareness campaigns for young craftsmen on the advantages of the bamboo craft sector;more training campaigns for craftsmen on modern bamboo processing techniques;and the establishment of a well-developed and sustainable bamboo-based craft sector.
文摘离线强化学习(Offline RL)定义了从固定批次的数据集中学习的任务,能够规避与环境交互的风险,提高学习的效率与稳定性。其中优势加权行动者-评论家算法提出了一种将样本高效动态规划与最大似然策略更新相结合的方法,在利用大量离线数据的同时,快速执行在线精细化策略的调整。但是该算法使用随机经验回放机制,同时行动者-评论家模型只采用一套行动者,数据采样与回放不平衡。针对以上问题,提出一种基于策略蒸馏并进行数据经验优选回放的优势加权双行动者-评论家算法(Advantage Weighted Double Actors-Critics Based on Policy Distillation with Data Experience Optimization and Replay,DOR-PDAWAC),该算法采用偏好新经验并重复回放新旧经验的机制,利用双行动者增加探索,并运用基于策略蒸馏的主从框架,将行动者分为主行为者和从行为者,提升协作效率。将所提算法应用到通用D4RL数据集中的MuJoCo任务上进行消融实验与对比实验,结果表明,其学习效率等均获得了更优的表现。
文摘深度强化学习在训练过程中会探索大量环境样本,造成算法收敛时间过长,而重用或传输来自先前任务(源任务)学习的知识,对算法在新任务(目标任务)的学习具有提高算法收敛速度的潜力。为了提高算法学习效率,提出一种双Q网络学习的迁移强化学习算法,其基于actor-critic框架迁移源任务最优值函数的知识,使目标任务中值函数网络对策略作出更准确的评价,引导策略快速向最优策略方向更新。将该算法用于Open AI Gym以及在三维空间机械臂到达目标物位置的实验中,相比于常规深度强化学习算法取得了更好的效果,实验证明提出的双Q网络学习的迁移强化学习算法具有较快的收敛速度,并且在训练过程中算法探索更加稳定。
