Please wait a minute...
Most Download
Most Cited
Links
Visited
Total visitors:
Visitors of today:
Now online:

Current Issue

    20 September 2020, Volume 2 Issue 3
    Review Intelligence
    AI in game intelligence—from multi-role game to parallel game
    Yu SHEN,Jinpeng HAN,Lingxi LI,Fei-Yue WANG
    2020, 2(3):  205-213.  doi:10.11959/j.issn.2096-6652.202023
    Asbtract ( 345 )   HTML ( 53)   PDF (6641KB) ( 648 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    The domestic and overseas research progress of artificial intelligence technology in the field of games was summarized and the significance of the research progress in the field of games for real life was analyzed.In view of the gap between simulation and reality in model based methods and the lack of generality of the model-based approach in reinforcement learning,the idea and method of parallel game were put forward,and the advance of parallel game in solving the existing problems of single-role game and multi-role game was introduced.The parallel game method will be the cornerstone of the general artificial intelligence.

    Surveys and Prospectives
    Intelligent innovative regulatory tools on financial technology:concept,platform framework,and prospects
    Hongfeng CHAI,Shuai WANG,Xiaojun TU,Quan SUN,Xiaofeng MA,Jie WU,Hua CAI,Xiaolong ZHENG,Fei-Yue WANG
    2020, 2(3):  214-226.  doi:10.11959/j.issn.2096-6652.202024
    Asbtract ( 278 )   HTML ( 31)   PDF (2729KB) ( 304 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    In view of the complex and severe challenges faced by financial technology regulation and the lag behind situation of the existing regulatory sandbox which is highly dependent on manual operation,the important feature and development trend of the Chinese “innovative regulatory tools on financial technology” lie in “intelligence” were pointed out.A systematic platform framework of “intelligent innovative regulatory tools on financial technology” was proposed,that is,use modern means of science and technology to realize the full life cycle supervision of financial technology innovative applications.The elements of innovative regulatory tools were elaborated,including the underlying blockchain,public opinion monitoring platform,explainable artificial intelligence based on knowledge graphs,and presented the key technologies involved.The new iterative methodology used in the development of the innovative regulatory tools platform was introduced,i.e.,the spiral development model and iterative design driven by the hierarchy of needs.Towards the end,some thoughts and prospects on the proposed innovative regulatory tools were put forward,and the regulatory tools are expected to create a set of regulatory technology infrastructure to serve regulatory authorities and financial institutions were pointed out.

    Regular Papers
    A survey of UAV-based edge intelligent computing
    Chao DONG,Yun SHEN,Yuben QU
    2020, 2(3):  227-239.  doi:10.11959/j.issn.2096-6652.202025
    Asbtract ( 999 )   HTML ( 125)   PDF (1089KB) ( 1356 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    Edge intelligent computing refers to the offloading of computationally intensive tasks generated by user nodes to edge servers with stronger computing capabilities for processing.Unmanned aerial vehicle (UAV)-based edge intelligent computing combines intelligent drone platforms on this basis and utilizes them with the advantages of strong mobility and easy deployment,it can provide edge computing services for ground user equipment more quickly and flexibly.At the same time,drones can also be used as user nodes to off load their computationally intensive tasks to the ground edge server for execution.Aiming at two different scenarios of UAV as a user node or an edge server,the current research on edge intelligent computing based on UAV is classified and summarized according to different optimization goals such as minimizing energy consumption,minimizing delay and maximizing utility,and the next research direction is considered and prospected.

    Classification of motor imagery signals using noise-assisted fast multivariate empirical mode decomposition
    Qian ZHENG,Dan QIAO,Xun LANG,Lei XIE,Dongliu Li,Qibing Wang,Hongye SU
    2020, 2(3):  240-250.  doi:10.11959/j.issn.2096-6652.202026
    Asbtract ( 153 )   HTML ( 20)   PDF (2764KB) ( 202 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    The brain-computer interface is an emerging technology,which can analyze the collected motor imagery signals to control the external auxiliary equipment.A new method based on the noise-assisted fast multivariate empirical mode decomposition (NA-FMEMD) algorithm was proposed for electroencephalogram signal feature extraction and classification.The method outperformed state-of-the-art methods based on noise-assisted multivariate empirical mode decomposition in not only computational efficiency but also classification accuracy.Firstly,all multivariate intrinsic mode functions and trend signals were obtained by the NA-FMEMD.Secondly,the multivariate signals with specific frequency bands were selected by computing their average frequencies.Thirdly,the common spatial pattern was applied to extract features.Finally,the feature vectors were classified using a support vector machine.Simulation data and BCI Competition IV data are used to verify the effectiveness and advantage of the new method,and the method is compared with noise-assisted multivariate empirical mode decomposition.

    Research on stochastic game and optimization in complex microgrid control
    Hong ZHOU,Ruotian YAO,Chang YU,Zhongcheng LEI
    2020, 2(3):  251-260.  doi:10.11959/j.issn.2096-6652.202027
    Asbtract ( 298 )   HTML ( 20)   PDF (3666KB) ( 206 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    A microgrid which integrated by a large scale of distributed generators,storage equipment and dynamic loads has stochastic characteristics in both the power generation and the demand-side.The purpose was to discuss various decision-making and optimal control approaches in the integrated energy microgrid based on a stochastic game.Firstly,the characteristic and architecture of the future complex microgrid were investigated.Secondly,the form and interact of the group game were explained in details,respectively.Then,the stochastic characteristics of different players and the topology that affecting the game on complex microgrids were presented.Finally,a stochastic game optimization approach and its control architecture was proposed.By discussing the key issues of microgrid synchronization,stability and game optimization in different perspectives,the new ideas for promoting clean energy consumption,supply-demand side interaction and maintaining the stability of power grids were provided.

    CFNN-based online control for dissolved oxygen concentration of wastewater treatment processes
    Limin QUAN,Cuili YANG,Junfei QIAO
    2020, 2(3):  261-267.  doi:10.11959/j.issn.2096-6652.202028
    Asbtract ( 145 )   HTML ( 15)   PDF (1110KB) ( 168 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    Due to the frequent disturbance in flow and load,as well as the large uncertainty in the wastewater treatment processes,it is difficult to control the dissolved oxygen accurately and in real-time.To improve the accuracy and robustness of the controller,an online control method of dissolved oxygen concentration using the correntropy based fuzzy neural network (CFNN) was proposed.First,the performance index was established based on the correntropy of tracking errors to suppress large outliers in the process.Then,the parameters of controller were updated by the online gradient descent algorithm.Moreover,the stability of the control system was analyzed.Finally,the experiments were carried out based on the Benchmark Simulation Model No.1 (BSM1).The results prove that the CFNN controller performs better than the mean square error based neural network controller in accuracy and model stability.

    A robot sorting method based on deep learning
    Sijia TIAN,Qiang GU,Rong HU,Ruige LI,Dingxin HE
    2020, 2(3):  268-274.  doi:10.11959/j.issn.2096-6652.202029
    Asbtract ( 230 )   HTML ( 25)   PDF (9369KB) ( 295 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    A fast robot sorting method based on lightweight convolutional neural network was proposed to improve the recognition speed and environmental adaptability,especially for sorting complex objects.Firstly,the MobileNet-SSD algorithm was used to detect and classify the objects based on lightweight convolutional neural network.Secondly,image preprocessing and edge extraction were used to revise the object locations according to the above object detection results.The sorting experiments on PROBOT Anno robot arm show that the proposed method can achieve fast detection and location for complex objects.Compared with traditional image processing methods,the proposed method is robust to the diversity of target morphology and environment.

    Cloud native robot system based on edge computing
    Dawei WANG,Zhuo WANG,Peng WANG,Zhigang WANG,Haitao WANG
    2020, 2(3):  275-283.  doi:10.11959/j.issn.2096-6652.202030
    Asbtract ( 439 )   HTML ( 45)   PDF (2651KB) ( 540 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    With the rapid development of artificial intelligence,the global robot market continues to grow quickly,and the capabilities of robots have evolved from performing fixed operations to the ability to autonomously sense,understand and make decisions.However,to achieve large-scale application of robots,robots need to have powerful computing capabilities and low deployment costs under the constraints of limited power consumption.Using edge computing to provide more cost-effective services,enhance the computing power of the robot body,and achieve large-scale deployment is the key to achieving this goal.The challenges faced by robot systems with the edge enhancement were analyzed,the concept of cloud-native robot systems based on edge computing was proposed,and four feasible technical solutions for implementing the system were discusses.The cloud-native robot system is the inevitable direction for the development of robot systems from intelligent systems based on robot ontology to cloud-edge-end fusion multi-robot collaborative intelligent systems and the key technology for promoting the large-scale application of robots.

    Research on capsule network-based for aspect-level sentiment classification
    Zhidong XU,Bingyang CHEN,Xiao WANG,Weishan ZHANG
    2020, 2(3):  284-292.  doi:10.11959/j.issn.2096-6652.202031
    Asbtract ( 153 )   HTML ( 31)   PDF (915KB) ( 254 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    Considering the difficulty of judging the mixed multiple sentimental polarities in a text,aspect-level sentiment analysis has become a hot research topic.Multiple sentiments of different targets when expressing multi-faceted in a sentence,it will cause problems such as feature overlap,which will have a negative impact on text sentiment classification.A capsule network-based model for aspect-level sentiment classification (SCACaps) was proposed.Sequential convolution was used to extract the features of context and aspect words separately,and an interactive attention mechanism was introduced to reduce the mutual influence on each other,and then the text feature representation was transmitted into the capsule network after reconstruction.The routing algorithm was optimized by introducing high-level capsule coefficients between the capsule layers,and the global parameters were shared in the entire iterative update process to save relatively complete text feature information.By comparing with multiple models,the SCACaps model has the best classification effect,and the SCACaps model also performs better in small sample learning.

    Parallel control and digital twins:control theory revisited and reshaped
    Fei-Yue WANG
    2020, 2(3):  293-300.  doi:10.11959/j.issn.2096-6652.202032
    Asbtract ( 522 )   HTML ( 65)   PDF (1524KB) ( 546 )   Knowledge map   
    Figures and Tables | References | Related Articles | Metrics

    After a brief discussion of the origin and development of classic control,intelligent control,and parallel control,the mathematical equations for a class of parallel control systems was presented for problems with high precision digital twin models.The key idea was modeling the time derivative of system control instead of system control itself,which leads to differential equation based control rather than algebraic relationship based control,made the system and its control symmetrical in both form and content mathematically,thus provided the foundation for implementing human-like control and intelligent control.Same idea was also applied to system output equations,and proposed a new way for active sensing and design of parallel sensors.Preliminary results along those lines are illustrated and discussed.