Abstract
In this paper, we propose a learning rule based on a back-propagation (BP) algorithm that can be applied to a hardware-based deep neural network using electronic devices that exhibit discrete and limited conductance characteristics. This adaptive learning rule, which enables forward, backward propagation, as well as weight updates in hardware, is helpful during the implementation of power-efficient and high-speed deep neural networks. In simulations using a three-layer perceptron network, we evaluate the learning performance according to various conductance responses of electronic synapse devices and weight-updating methods. It is shown that the learning accuracy is comparable to that obtained when using a software-based BP algorithm when the electronic synapse device has a linear conductance response with a high dynamic range. Furthermore, the proposed unidirectional weight-updating method is suitable for electronic synapse devices which have nonlinear and finite conductance responses. Because this weight-updating method can compensate the demerit of asymmetric weight updates, we can obtain better accuracy compared to other methods. This adaptive learning rule, which can be applied to full hardware implementation, can also compensate the degradation of learning accuracy due to the probable device-to-device variation in an actual electronic synapse device.
Similar content being viewed by others
References
Poon CS, Zhou K (2011) Neuromorphic silicon neurons and large-scale neural networks: challenges and opportunities. Front Neurosci 5:108. https://doi.org/10.3389/fnins.2011.00108
Suri M, Bichler O, Querlioz D, Cueto O, Perniola L, Sousa V, Vuillaume D, Gamrat C, DeSalvo B (2011) Phase change memory as synapse for ultra-dense neuromorphic systems: application to complex visual pattern extraction. IEEE Electron Devices Meeting. https://doi.org/10.1109/IEDM.2011.6131488
Wright CD, Liu Y, Kohary KI, Aziz MM, Hicken RJ (2011) Arithmetic and biologically-inspired computing using phase-change materials. Adv Mater 23:3408–3413. https://doi.org/10.1002/adma.201101060
Kuzum D, Jeyasingh RG, Lee B, Wong HS (2012) Nanoelectronic programmable synapses based on phase change materials for brain-inspired computing. Nano Lett 12:2179–2186. https://doi.org/10.1021/nl201040y
Jo SH, Chang T, Ebong I, Bhadviya BB, Mazumder P, Lu W (2010) Nanoscale memristor device as synapse in neuromorphic systems. Nano Lett 10:1297–1301. https://doi.org/10.1021/nl904092h
Ohno T, Hasegawa T, Tsuruoka T, Terabe K, Gimzewski JK, Aono M (2011) Short-term plasticity and long-term potentiation mimicked in single inorganic synapses. Nat Mater 10:591–595. https://doi.org/10.1038/nmat3054
Wu Y, Yu S, Wong HS (2012) AlOx-based resistive switching device with gradual resistance modulation for neuromorphic device application. IEEE Memory Workshop. https://doi.org/10.1109/IMW.2012.6213663
Yu S, Gao B, Fang Z (2012) A neuromorphic visual system using RRAM synaptic devices with sub-pJ energy and tolerance to variability: experimental characterization and large-scale modeling. IEEE Electron Devices Meeting. https://doi.org/10.1109/IEDM.2012.6479018
Chanthbouala A, Garcia V, Cherifi RO, Bouzehouane K, Fusil S, Moya X, Xavier S, Yamada H, Deranlot C, Mathur ND, Bibes M, Barthelemy A, Grollier J (2012) A ferroelectric memristor. Nat Mater 11:860–864. https://doi.org/10.1038/nmat3415
Diorio C, Hasler P, Minch A, Mead CA (1996) A single-transistor silicon synapse. IEEE Trans Electron Devices 43:1972–1980. https://doi.org/10.1109/16.543035
Ziegler M, Oberländer M, Schroeder D, Krautschneider WH, Kohlstedt H (2012) Memristive operation mode of floating gate transistors: a two-terminal MemFlash-cell. Appl Phys Lett 101:263504. https://doi.org/10.1063/1.4773300
Kim H, Park J, Kwon M-W, Lee J-H, Park B-G (2016) Silicon-based floating-body synaptic transistor with frequency-dependent short- and long-term memories. IEEE Electron Device Lett 37:249–252. https://doi.org/10.1109/led.2016.2521863
Bi G-q, Poo M-m (1998) Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type. J Neurosci 18:10464. https://www.ncbi.nlm.nih.gov/pubmed/9852584
Kuzum D, Yu S, Wong HS (2013) Synaptic electronics: materials, devices and applications. Nanotechnology 24:382001. https://doi.org/10.1088/0957-4484/24/38/382001
Burr GW, Shelby RM, di Nolfo C, Jang J-W, Shenoy RS, Narayanan P, Virwani K, Giacometti EU, Kurdi B, Hwang H (2014) Experimental demonstration and tolerancing of a large-scale neural network (165,000 synapses), using phase-change memory as the synaptic weight element. IEEE Electron Devices Meeting. https://doi.org/10.1109/iedm.2014.7047135
Nair MV, Dudek P (2015) Gradient-descent-based learning in memristive crossbar arrays. IEEE Int Joint Conf Neural Netw. https://doi.org/10.1109/IJCNN.2015.7280658
Prezioso M, Merrikh-Bayat F, Hoskins BD, Adam GC, Likharev KK, Strukov DB (2015) Training and operation of an integrated neuromorphic network based on metal-oxide memristors. Nature 521:61–64. https://doi.org/10.1038/nature14441
Merrikh-Bayat F, Guo X, Om’mani HA, Do N, Likharev KK, Strukov DB (2015) Redesigning commercial floating-gate memory for analog computing applications. IEEE Int Symp Circuits Syst. https://doi.org/10.1109/ISCAS.2015.7169048
Rumelhart DE, Hinton GE, Williams RJ (1986) Learning internal representations by error propagation. In: Parallel distributed processing: explorations in macrostructure of cognition, vol I. Badford, Cambridge. https://dl.acm.org/citation.cfm?id=104293
Krizhevsky A, Sutskever I, and Hinton GE (2012) ImageNet classification with deep convolutional neural networks. In: Advances in neural information processing systems, pp 1–9. https://dl.acm.org/citation.cfm?id=2999257
He K, Zhang X, Ren S, Sun J (2015) Deep residual learning for image recognition. CoRR arXiv:abs/1512.03385
Cho K, van Merrienboer B, Gülçehre Ç, Bahdanau D, Bougares F, Schwenk H, Bengio Y (2014) Learning phrase representations using RNN encoder–decoder for statistical machine translation. arXiv:1406.1078
Greff K, Srivastava RK, Koutnik J, Steunebrink BR, Schmidhuber J (2017) LSTM: a search space odyssey. IEEE Trans Neural Netw Learn Syst 28(10):2222–2232. https://doi.org/10.1109/TNNLS.2016.2582924
Deng Y, Ren Z, Kong Y, Bao F, Dai Q (2017) A hierarchical fused fuzzy deep neural network for data classification. IEEE Trans Fuzzy Syst 25:1006–1012. https://doi.org/10.1109/tfuzz.2016.2574915
Deng Y, Bao F, Kong Y, Ren Z, Dai Q (2017) Deep direct reinforcement learning for financial signal representation and trading. IEEE Trans Neural Netw Learn Syst 28:653–664. https://doi.org/10.1109/TNNLS.2016.2522401
Merkel C, Hasan R, Soures N, Kudithipudi D, Taha T, Agarwal S, Marinella M (2016) Neuromemristive systems: boosting efficiency through brain-inspired computing. Computer 49:56–64. https://doi.org/10.1109/MC.2016.312
Neftci EO, Augustine C, Paul S, Detorakis G (2017) Event-driven random back-propagation: enabling neuromorphic deep learning machines. Front Neurosci 11:324. https://doi.org/10.3389/fnins.2017.00324
Merrikh Bayat F, Bagheri Shouraki S, Esmaili Paeen Afrakoti I (2013) Bottleneck of using a single memristive device as a synapse. Neurocomputing 115:166–168. https://doi.org/10.1016/j.neucom.2012.12.027
Jang J-W, Park S, Burr GW, Hwang H, Jeong Y-H (2015) Optimization of conductance change in PrCaMnO based synaptic devices for neuromorphic systems. IEEE Electron Device Lett 36:457–459. https://doi.org/10.1109/LED.2015.2418342
Woo J, Moon K, Song J, Kwak M, Park J, Hwang H (2016) Optimized programming scheme enabling linear potentiation in filamentary HfO2 RRAM synapse for neuromorphic systems. IEEE Trans Electron Devices 63:5064–5067. https://doi.org/10.1109/ted.2016.2615648
Fuller EJ, Gabaly FE, Leonard F, Agarwal S, Plimpton SJ, Jacobs-Gedrim RB, James CD, Marinella MJ, Talin AA (2017) Li-ion synaptic transistor for low power analog computing. Adv Mater 29(4):1604310. https://doi.org/10.1002/adma.201604310
Bae J-H, Lim S, Park B-G, Lee J-H (2017) High-density and near-linear synaptic device based on a reconfigurable gated Schottky diode. IEEE Electron Device Lett 38:1153–1156. https://doi.org/10.1109/LED.2017.2713460
Gokmen T, Vlasov Y (2016) Acceleration of deep neural network training with resistive cross-point devices: design considerations. Front Neurosci 10:333. https://doi.org/10.3389/fnins.2016.00333
Fumarola A, Narayana P, Sanches LL, Sidler S, Jang J, Moon K, Shelby RM, Hwang H, Burr GW (2016) Accelerating machine learning with non-volatile memory: exploring device and circuit tradeoffs. IEEE Int Conf Rebooting Comput. https://doi.org/10.1109/ICRC.2016.7738684
Binas J, Neil D, Indiveri G, Liu S-C, Pfeiffer M (2016) Precise deep neural network computation on imprecise low-power analog hardware. arXiv preprint. arXiv:1606.07786
Schiffmann W, Joost M, Werner R (1994) Optimization of the backpropagation algorithm for training multilayer perceptrons. Technical report, University of Koblenz, Institute of Physics, Rheinau. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.53.6869&rep=rep1&type=pdf
Bichler O, Suri M, Querlioz D, Vuillaume D, DeSalvo B, Gamrat C (2012) Visual pattern extraction using energy-efficient “2-PCM Synapse” neuromorphic architecture. IEEE Trans Electron Devices 59:2206–2214. https://doi.org/10.1109/ted.2012.2197951
MATLAB and Statistics Toolbox Release 2016a. The MathWorks, Inc., Natick
Lecun Y, Bottou L, Bengio Y, Haffner P (1998) Gradient-based learning applied to document recognition. Proc IEEE 86:2278–2324. https://doi.org/10.1109/5.726791
Querlioz D, Bichler O, Dollfus P, Gamrat C (2013) Immunity to device variations in a spiking neural network with memristive nanodevices. IEEE Trans Nanotechnol 12:288–295. https://doi.org/10.1109/TNANO.2013.2250995
Bengio Y (2009) Learning deep architectures for AI. Found Trends Mach Learn 2:1–127. https://doi.org/10.1561/2200000006
Bengio Y, Lamblin P, Popovici D, Larochelle H (2006) Greedy layer-wise training of deep networks. Adv Neural Inf Process Syst 153–160. https://dl.acm.org/citation.cfm?id=2976476
Srivastava N, Hinton GE, Krizhevsky A, Sutskever I, Salakhutdinov R (2014) Dropout: a simple way to prevent neural networks from overfitting. The J Mach Learn Res 15:1929–1958. http://dl.acm.org/citation.cfm?id=2670313
Acknowledgements
This work was supported by the Korea Institute of Science and Technology (KIST) Institutional Program (Project No. 2E27810-18-P040) and the Brain Korea 21 Plus Project in 2018.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Rights and permissions
About this article
Cite this article
Lim, S., Bae, JH., Eum, JH. et al. Adaptive learning rule for hardware-based deep neural networks using electronic synapse devices. Neural Comput & Applic 31, 8101–8116 (2019). https://doi.org/10.1007/s00521-018-3659-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-018-3659-y