logo

SCIENTIA SINICA Informationis, Volume 48, Issue 7: 824-840(2018) https://doi.org/10.1360/N112017-00303

Backtracking analysis approach for effectiveness of air defense operation system of systems based on force-sparsed stacked-autoencoding neural networks

More info
  • ReceivedDec 31, 2017
  • AcceptedJan 28, 2018
  • PublishedJul 19, 2018

Abstract

To reduce the difficulty of traditional data-mining methods in deep analysis of emerging capacity index mechanisms for a complex air defense system of systems (ADSOS), a novel backtracking analysis approach based on force-sparsed stacked-autoencoding (FS-SAE) neural networks is proposed. Using the method of big-data correlation analysis, community structure analysis in complex networks, and principle component analysis based on the mission task, the networked structure of an index system for ADSOS, which was a relatively complete architecture with a well-defined meaning, was obtained. An FS-SAE backtracking analysis model was build based on heuristic knowledge from the obtained networked index system structure. The emerging relations between the capacity indices of an ADSOS were formalized. Then, the formation mechanism and contribution rate of capacity indices were studied, and the validity of this approach was validated by the simulation data. The experimental results show that formalized presentation for the emergence process of capacity indices of ADSOS based on the proposed model not only reflects the complexity characteristics of nonlinearity and uncertainty in the emergence process, but also give a general defined meaning for the index structure of the ADSOS. It provides a feasible method for commanders to deeply understand, manage, and control the complex operation system of systems.


Funded by

国家自然科学基金(61273189)

国家自然科学基金(71401168)

国家自然科学基金(61174156)

国家自然科学基金(61403401)

国家自然科学基金(61374179)

军民共用重大研究计划联合基金(U1435218)


References

[1] Hu X F. On War System Engineering: Methodology Towards to Information Age's War. Beijing: University of National Defense Press, 2013. 200--413. Google Scholar

[2] Hu X F, Rong M. Joint operation planning systems: issues and prospects. J Command Control, 2017, 3: 273--280. Google Scholar

[3] Yao X Y, Guo S M, Hu X F, et al. Construction for air defense system of systems based on MINE-FNA assembled algorithm. Command Control Simul, 2016, 1: 94--100. Google Scholar

[4] Tolk A. The next generation of modeling & simulation: integrating big data and deep learning. In: Proceedings of the Conference on Summer Computer Simulation, Chicago, 2015. Google Scholar

[5] Tannahill B K, Jamshidi M. System of Systems and big data analytics — bridging the gap. Comput Electr Eng, 2014, 40: 2-15 CrossRef Google Scholar

[6] Hu X F, He X Y, Xu X L. Simulation in the big data era — review of new ideas and new theories in the 81st Academic Salon of China Association for Science and Technology. Sci Sin Inform, 2014, 44: 676--692. Google Scholar

[7] Zhang B, Hu X F, Wang Y, et al. Rudimental research of analysis of system of systems exceptional state. J Syst Simul, 2011, 23: 2494--2499. Google Scholar

[8] Wu W F, Hu X F. Capacity evaluation of networked operational system of systems based on big data. Mil Oper Res Syst Eng, 2016, 30: 26--32. Google Scholar

[9] Wu W F, Hu X F, Guo S M, et al. Analysis of combat SoS coordination based on multi-layered temporal networks. Complex Syst Complex Sci, 2017, 14: 1--10. Google Scholar

[10] Yao X Y, Guo S M, Hu X F, et al. Design evaluation indices for air defense systems based on RF-GLR integrated machine learning algorithms. J Command Control. 2015, 1: 269--277. Google Scholar

[11] Hu X F, Zhang Y, Li R J, et al. Capability evaluating problem of networking SOS. Syst Eng Theory Pract, 2015, 35: 1317--1323. Google Scholar

[12] Ding J F, Si G Y, Yang G Q. Visualization analysis of the capability of weapon system of systems for multi-dimensional indicators. J Syst Eng Electron, 2017, 28: 292-300 CrossRef Google Scholar

[13] Silver D, Huang A, Maddison C J. Mastering the game of Go with deep neural networks and tree search. Nature, 2016, 529: 484-489 CrossRef PubMed ADS Google Scholar

[14] Brown N, Sandholm T. Superhuman AI for heads-up no-limit poker: libratus beats top professionals. Science, 2018, 359: 418-424 CrossRef PubMed Google Scholar

[15] Elaraby N M, Elmogy M, Barakat S. Deep learning: effective tool for big data analytics. Int J Comput Sci Eng (IJCSE), 2016, 5: 254--262. Google Scholar

[16] Hu X F, Guo S M, He X Y. Challenges to intelligent command information system: reason and revelation on deep green plan and AlphaGo. Command Inf Syst Technol, 2016, 7: 1--7. Google Scholar

[17] Guo S M, He X Y, Hu X F, et al. Challenges and trends in intelligent military information system. Control Theory Appl, 2016, 33: 1562--1571. Google Scholar

[18] Tao J Y, Wu L, Hu X F. Principle analysis on AlphaGo and perspective in military application of artificial intelligence. J Command Control, 2016, 2: 114--120. Google Scholar

[19] Goodfellow I, Bengio Y, Courville A. Deep Learning. Cambridge: MIT press, 2016. Google Scholar

[20] Lee M, Bressler S, Kozma R. Advances in cognitive engineering using neural networks. Neur Netw, 2017, 92: 1-2 CrossRef PubMed Google Scholar

[21] Chen Q J, Song X, Yamada H, et al. Learning deep representation from big and heterogeneous data for traffic accident inference. In: Proceedings of the 30th AAAI Conference on Artificial Intelligenc, Phoenix, 2016. 338--344. Google Scholar

[22] Angermueller C, P?rnamaa T, Parts L. Deep learning for computational biology. Mol Syst Biol, 2016, 12: 878 CrossRef Google Scholar

[23] Zhang L, Liu Z, Zhang J Q, et al. Reduction method of weapon system of systems assessment index system based on autoencoder. J Cent South Univ (Sci Technol), 2013, 44: 4130--4137. Google Scholar

[24] Ou W, Liu S J, He X Y, et al. Tactical intention recognition algorithm based on encoded temporal features. Command Control Simul, 2016, 38: 36--41. Google Scholar

[25] Luong M T, Pham H, Manning C D. Effective approaches to attention-based neural machine translation. In: Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, Lisbon, 2015. 1412--1421. Google Scholar

[26] Sukhbaatar S, Weston J, Fergus R. End-to-end memory networks. In: Proceedings of Advances in Neural Information Processing Systems, Montreal, 2015. 2440--2448. Google Scholar

[27] Mnih V, Kavukcuoglu K, Silver D. Human-level control through deep reinforcement learning. Nature, 2015, 518: 529-533 CrossRef PubMed ADS Google Scholar

[28] Reshef D N, Reshef Y A, Finucane H K. Detecting novel associations in large data sets. Science, 2011, 334: 1518-1524 CrossRef PubMed ADS Google Scholar

[29] Girvan M, Newman M E J. Community structure in social and biological networks. Proc Natl Acad Sci USA, 2002, 99: 7821-7826 CrossRef PubMed ADS Google Scholar

[30] Martinez A M, Kak A C. PCA versus LDA. IEEE Trans Pattern Anal Mach Intel, 2001, 23: 228-233 CrossRef Google Scholar

[31] Meng L H, Ding S F, Zhang N, et al. Research of stacked denoising sparse autoencoder. In: Neural Computing and Applications. Berlin: Springer, 2016. 1--18. Google Scholar

[32] Bengio Y. Deep Learning Tutorial. 2014. http://deeplearning.net. Google Scholar

[33] Zhou X, Guo J Q, Wang S L. Motion recognition by using a stacked autoencoder-based deep learning algorithm with smart phones. In: Proceedings of International Conference on Wireless Algorithms, Systems, and Applications, Qufu, 2015. 778--787. Google Scholar

[34] Wu W F, Hu X F, Guo S M, et al. Methods of analyzing combat SoS coordination pattern based on temporal motif. In: Proceedings of Asian Simulation Conference, Beijing, 2016. 544--554. Google Scholar

  • Figure 1

    (Color online) The idea of backtracking analysis approach based on FS-SAE

  • Figure 2

    (Color online) The design route for the structure of FS-SAE model. (a) Initial index sets; (b) initial index networks; (c) function indexes for SOS; (d) eigen indexes of community; (e) multilayered structure

  • Figure 3

    The flowchart of GN algorithm

  • Figure 4

    (Color online) Scenario setting and the construction of the networked index system of ADSOS. (a) Illustration of scenario setting; (b) initial index networks; (c) index network communities; (d) multilayer networked index system

  • Figure 5

    (Color online) Convergence comparison of different algorithms

  • Figure 6

    (Color online) The distribution of function index expectation of ADSOS. (a) The probability of function index expectation; (b) the range and the average value of function index expectation

  • 1   Table 1MICs between the initial indexes
    Index $X$1 $X$2 $X$3 $X$4 $X$5 $X$6
    $X$2 0.22279
    $X$3 0.24872 0.87671
    $X$4 0.23126 0.51958 0.71057
    $X$5 0.24557 0.51958 0.55146 0.43817
    $X$6 0.95095 0.40619 0.24412 0.50126 0.99107
  • 2   Table 2The accuracy of FS-SAE models for the mission
    Model Training set accuracy (%) Testing set accuracy (%)
    FS-SAE 95 82
    Sparsed-SAE 84 73

Copyright 2019 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有

京ICP备18024590号-1