A spintronic memristive circuit on the optimized RBF-MLP neural network
Yuan Ge(葛源)1, Jie Li(李杰)1, Wenwu Jiang(蒋文武)1, Lidan Wang(王丽丹)1,2,3,4, and Shukai Duan(段书凯)1,2,3,4,†
1 School of Artificial Intelligence, Southwest University, Chongqing 400715, China; 2 Chongqing Brain Science Collaborative Innovation Center, Chongqing 400715, China; 3 Brain-inspired Computing and Intelligent Control of Chongqing Key Laboratory, Chongqing 400715, China; 4 National&Local Joint Engineering Laboratory of Intelligent Transmission and Control Technology, Chongqing 400715, China
Abstract A radial basis function network (RBF) has excellent generalization ability and approximation accuracy when its parameters are set appropriately. However, when relying only on traditional methods, it is difficult to obtain optimal network parameters and construct a stable model as well. In view of this, a novel radial basis neural network (RBF-MLP) is proposed in this article. By connecting two networks to work cooperatively, the RBF's parameters can be adjusted adaptively by the structure of the multi-layer perceptron (MLP) to realize the effect of the backpropagation updating error. Furthermore, a genetic algorithm is used to optimize the network's hidden layer to confirm the optimal neurons (basis function) number automatically. In addition, a memristive circuit model is proposed to realize the neural network's operation based on the characteristics of spin memristors. It is verified that the network can adaptively construct a network model with outstanding robustness and can stably achieve 98.33% accuracy in the processing of the Modified National Institute of Standards and Technology (MNIST) dataset classification task. The experimental results show that the method has considerable application value.
Yuan Ge(葛源), Jie Li(李杰), Wenwu Jiang(蒋文武), Lidan Wang(王丽丹), and Shukai Duan(段书凯) A spintronic memristive circuit on the optimized RBF-MLP neural network 2022 Chin. Phys. B 31 110702
[1] Javed K, Jun S, Markus R, Lao S, Marc L, Frank W, Frank B, Manfred S, Cristina A, Carsten P and Paul S M 2001 Nat. Med. 7 673 [2] Giuseppe C and Matthias T 2017 Science355 602 [3] Gardner M W 1998 Atmos. Environ.32 2627 [4] Robert H N 1988 Elsevier65 93 [5] Widrow B and Lehr M A 2002 IEEE Xplore78 1415 [6] Huang D S 1999 World Scientific13 1083 [7] Lee C C and Chung P C 1999 IEEE Transactions on Systems Man & Cybernetics Part B 29 674 [8] Yue H, Zhang H J and Chai T Y 2006 Control Engineering of China [9] Schwenker F, Kestler H A and Palm G 2001 Neural Network14 439 [10] Purnawansyah B N and Haviluddin H 2016 Makassar International Conference on Electrical Engineering & Informatics 32 124 [11] Li T S, Duan S K, Liu J, Wang L D and Huang T W 2016 IEEE Trans. Syst. Man Cybern. Syst.46 582 [12] Young S H and Sung Y B 1997 Neural Networks the Official Journal of the International Neural Network Society10 1495 [13] Kanungo T, Mount D M, Netanyahu N S, Sliverman R and Wu A Y 2002 IEEE Transactions on Pattern Analysis & Machine Intelligence 24 881 [14] Tao X L and Michel H E 2008 International Society for Optics and Photonics5267 311 [15] Ye J, Ge L D and Wu Y X 2007 Acta Automatica Sinica33 652 [16] Pei X M, Jin X, Zheng C X and Bin G Y 2005 Advances in Natural Computation, pp. 1031-1034 [17] Wang X, Tian J and Wang M J 2010 International Conference on Intelligent Control and Information Processing [18] Han H G, Qiao J F and Bo Y C 2012 Acta Automatica Sinica38 1083 [19] Duan S K, Hu X F, Dong Z K and Wang L D 2015 IEEE Trans. Neural Netw. Learn. Syst.26 1202 [20] Klidbary S H and Shouraki S B 2018 Appl. Intell.48 4174 [21] Adhikari S P, Kim H, Budhathoki R K, Yang C J and Chua L 2015 IEEE Transactions on Circuits and Systems I: Regular Papers62 215 [22] Chua L 1971 IEEE Trans. Circuit Theory18 [23] Biolek D, Biolek Z and Biolkova V 2009 European Conference on Circuit Theory & Design [24] Kim K H, Gaba S, Wheeler D, Hussain T, Srinivasa T and Lu W 2012 Nano Lett.12 389 [25] Xia Q, Robinett W and Cumbie M W 2009 Nano Lett.9 3640 [26] Wang L, Yang C H, Wen J, Gai S and Peng Y X 2015 Journal of Materials Science Materials in Electronics26 4618 [27] Allan P 1999 Acta Numer.8 143 [28] Li Q D, Xiong Q Y, Ji S F, Yu Y, Wu C and Yi H L 2021 Neurocomputing431 7 [29] Nabney I T 2004 International Journal of Neural Systems14 201 [30] Zeiler M and Fergus R 2013 European Conference on Computer Vision 818 [31] Goldberg D E 1999 Optimization & Machine Learning [32] Kazarlis S A, Bakirtzis A G and Petridis V 1996 IEEE Trans. Power Syst.11 83 [33] Sripriya J, Ramalingam A and Rajeswari K 2015 International Conference on Innovations in Information [34] Deb K and Beyer H G 2001 Evolutionary Computation9 197 [35] Agrawal R B and Deb K 1994 Complex Systems9 115 [36] Yang J and Honavar V 2002 IEEE Intelligent Systems & Their Applications 13 44 [37] Rovithakis G A, Maniadakis M and Zervakis M 2019 IEEE Transactions Syst., Man, Cybern.34 695 [38] Fan D L, Sharad M and Roy K 2014 Nanotechnology13 574 [39] Kaslik E and Balint S 2008 Journal of Nonlinear Science18 415 [40] Xi Z and Panoutsos G 2019 International Conference on Intelligent Systems 1541 [41] Zadeh P H, Hosseini R and Sra S 2018 Cornell Computer Science [42] Nebti S and Boukerram A 2010 International Conference on Machine and Web Intelligence 464 [43] Zyarah A M, Ramesh A and Merkel C 2016 Machine Intelligence & Bio-inspired Computation 13 72 [44] Allred J M and Roy K 2019 Cornell Computer Science2
Altmetric calculates a score based on the online attention an article receives. Each coloured thread in the circle represents a different type of online attention. The number in the centre is the Altmetric score. Social media and mainstream news media are the main sources that calculate the score. Reference managers such as Mendeley are also tracked but do not contribute to the score. Older articles often score higher because they have had more time to get noticed. To account for this, Altmetric has included the context data for other articles of a similar age.
