Abstract

In recent years, artificial neural networks have become the flagship algorithm of artificial intelligence1. In these systems, neuron activation functions are static, and computing is achieved through standard arithmetic operations. By contrast, a prominent branch of neuroinspired computing embraces the dynamical nature of the brain and proposes to endow each component of a neural network with dynamical functionality, such as oscillations, and to rely on emergent physical phenomena, such as synchronization2,3,4,5,6, for solving complex problems with small networks7,8,9,10,11. This approach is especially interesting for hardware implementations, because emerging nanoelectronic devices can provide compact and energy-efficient nonlinear auto-oscillators that mimic the periodic spiking activity of biological neurons12,13,14,15,16. The dynamical couplings between oscillators can then be used to mediate the synaptic communication between the artificial neurons. One challenge for using nanodevices in this way is to achieve learning, which requires fine control and tuning of their coupled oscillations17; the dynamical features of nanodevices can be difficult to control and prone to noise and variability18. Here we show that the outstanding tunability of spintronic nano-oscillators—that is, the possibility of accurately controlling their frequency across a wide range, through electrical current and magnetic field—can be used to address this challenge. We successfully train a hardware network of four spin-torque nano-oscillators to recognize spoken vowels by tuning their frequencies according to an automatic real-time learning rule. We show that the high experimental recognition rates stem from the ability of these oscillators to synchronize. Our results demonstrate that non-trivial pattern classification tasks can be achieved with small hardware neural networks by endowing them with nonlinear dynamical features such as oscillations and synchronization.

Access optionsAccess options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Data availability

The datasets generated and analysed during this study are available from the corresponding authors on reasonable request.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  1. 1.

    Silver, D. et al. Mastering the game of Go without human knowledge. Nature 550, 354–359 (2017).

  2. 2.

    Borisyuk, R., Denham, M., Hoppensteadt, F., Kazanovich, Y. & Vinogradova, O. An oscillatory neural network model of sparse distributed memory and novelty detection. Biosystems 58, 265–272 (2000).

  3. 3.

    Jaeger, H. & Haas, H. Harnessing nonlinearity: predicting chaotic systems and saving energy in wireless communication. Science 304, 78–80 (2004).

  4. 4.

    Rabinovich, M., Huerta, R. & Laurent, G. Transient dynamics for neural processing. Science 321, 48–50 (2008).

  5. 5.

    Sussillo, D. Neural circuits as computational dynamical systems. Curr. Opin. Neurobiol. 25, 156–163 (2014).

  6. 6.

    Pikovsky, A. & Rosenblum, M. Dynamics of globally coupled oscillators: progress and perspectives. Chaos 25, 097616 (2015).

  7. 7.

    Kumar, S., Strachan, J. P. & Williams, R. S. Chaotic dynamics in nanoscale NbO2 Mott memristors for analogue computing. Nature 548, 318–321 (2017).

  8. 8.

    Csaba, G. & Porod, W. Computational study of spin-torque oscillator interactions for non-Boolean computing applications. IEEE Trans. Magn. 49, 4447–4451 (2013).

  9. 9.

    Yogendra, K., Fan, D., Jung, B. & Roy, K. Magnetic pattern recognition using injection-locked spin-torque nano-oscillators. IEEE Trans. Electron Dev. 63, 1674–1680 (2016).

  10. 10.

    Macià, F., Kent, A. D. & Hoppensteadt, F. C. Spin-wave interference patterns created by spin-torque nano-oscillators for memory and computation. Nanotechnology 22, 095301 (2011).

  11. 11.

    Fang, Y., Yashin, V. V., Levitan, S. P. & Balazs, A. C. Pattern recognition with “materials that compute”. Sci. Adv. 2, e1601114 (2016).

  12. 12.

    Pickett, M. D., Medeiros-Ribeiro, G. & Williams, R. S. A scalable neuristor built with Mott memristors. Nat. Mater. 12, 114–117 (2013).

  13. 13.

    Pufall, M. R. et al. Physical implementation of coherently coupled oscillator networks. IEEE J. Explor. Solid-State Comput. Devices Circuits 1, 76–84 (2015).

  14. 14.

    Sharma, A. A., Bain, J. A. & Weldon, J. A. Phase coupling and control of oxide-based oscillators for neuromorphic computing. IEEE J. Explor. Solid-State Comput. Devices Circuits 1, 58–66 (2015).

  15. 15.

    Grollier, J., Querlioz, D. & Stiles, M. D. Spintronic nanodevices for bioinspired computing. Proc. IEEE 104, 2024–2039 (2016).

  16. 16.

    Parihar, A., Shukla, N., Jerry, M., Datta, S. & Raychowdhury, A. Computational paradigms using oscillatory networks based on state-transition devices. In 2017 International Joint Conference on Neural Networks (IJCNN) 3415–3422 (IEEE, 2017).

  17. 17.

    Vassilieva, E., Pinto, G., de Barros, J. A. & Suppes, P. Learning pattern recognition through quasi-synchronization of phase oscillators. IEEE Trans. Neural Netw. 22, 84–95 (2011).

  18. 18.

    Vodenicarevic, D., Locatelli, N., Araujo, F. A., Grollier, J. & Querlioz, D. A nanotechnology-ready computing scheme based on a weakly coupled oscillator network. Sci. Rep. 7, 44772 (2017).

  19. 19.

    Locatelli, N., Cros, V. & Grollier, J. Spin-torque building blocks. Nat. Mater. 13, 11–20 (2014).

  20. 20.

    Slavin, A. & Tiberkevich, V. Nonlinear auto-oscillator theory of microwave generation by spin-polarized current. IEEE Trans. Magn. 45, 1875–1918 (2009).

  21. 21.

    Kaka, S. et al. Mutual phase-locking of microwave spin torque nano-oscillators. Nature 437, 389–392 (2005).

  22. 22.

    Mancoff, F. B., Rizzo, N. D., Engel, B. N. & Tehrani, S. Phase-locking in double-point-contact spin-transfer devices. Nature 437, 393–395 (2005).

  23. 23.

    Houshang, A. et al. Spin-wave-beam driven synchronization of nanocontact spin-torque oscillators. Nat. Nanotech. 11, 280–286 (2016).

  24. 24.

    Lebrun, R. et al. Mutual synchronization of spin torque nano-oscillators through a long-range and tunable electrical coupling scheme. Nat. Commun. 8, 15825 (2017).

  25. 25.

    Torrejon, J. et al. Neuromorphic computing with nanoscale spintronic oscillators. Nature 547, 428–431 (2017).

  26. 26.

    Tsunegi, S. et al. High emission power and Q factor in spin torque vortex oscillator consisting of FeB free layer. Appl. Phys. Express 7, 063009 (2014).

  27. 27.

    Romera, M. et al. Enhancing the injection locking range of spin torque oscillators through mutual coupling. Appl. Phys. Lett. 109, 252404 (2016).

  28. 28.

    Hillenbrand, J., Getty, L. A., Wheeler, K. & Clark, M. J. Acoustic characteristics of American English vowels. J. Acoust. Soc. Am. 97, 3099–3111 (1994).

  29. 29.

    Vodenicarevic, D., Locatelli, N., Grollier, J. & Querlioz, D. Synchronization detection in networks of coupled oscillators for pattern recognition. In 2016 International Joint Conference on Neural Networks (IJCNN) 2015–2022 (IEEE, 2016).

  30. 30.

    Fang, B. et al. Giant spin-torque diode sensitivity in the absence of bias magnetic field. Nat. Commun. 7, 11259 (2016).

  31. 31.

    Tulapurkar, A. A. et al. Spin-torque diode effect in magnetic tunnel junctions. Nature 438, 339–342 (2005).

  32. 32.

    Louis, S. et al. Low power microwave signal detection with a spin-torque nano-oscillator in the active self-oscillating regime. IEEE Trans. Magn. 53, 1–4 (2017).

  33. 33.

    Jouppi, N. P. et al. Datacenter performance analysis of a tensor processing unit. In Proc. 44th Annual International Symposium on Computer Architecture 1–12 (ACM, 2017).

  34. 34.

    Livi, P. & Indiveri, G. A current-mode conductance-based silicon neuron for address-event neuromorphic systems. In 2009 IEEE International Symposium on Circuits and Systems 2898–2901 (IEEE, 2009).

  35. 35.

    Qiao, N. & Indiveri, G. Scaling mixed-signal neuromorphic processors to 28 nm FD-SOI technologies. In 2016 IEEE Biomedical Circuits and Systems Conference (BioCAS) 552–555 (IEEE, 2016).

  36. 36.

    Wijekoon, J. H. B. & Dudek, P. Compact silicon neuron circuit with spiking and bursting behaviour. Neural Netw. 21, 524–534 (2008).

  37. 37.

    Tran, D. X. & Dang, T. T. An ultra-low power consumption and very compact 1.49 GHz CMOS voltage controlled ring oscillator. In 2014 International Conference on Advanced Technologies for Communications (ATC 2014) 239–244 (IEEE, 2014).

  38. 38.

    Tomita, Y. et al. An 8-to-16GHz 28nm CMOS clock distribution circuit based on mutual-injection-locked ring oscillators. In 2013 Symposium on VLSI Circuits C238–C239 (IEEE, 2013).

  39. 39.

    Gajek, M. et al. Spin torque switching of 20 nm magnetic tunnel junctions with perpendicular anisotropy. Appl. Phys. Lett. 100, 132408 (2012).

Download references

Acknowledgements

This work was supported by the European Research Council ERC under grant bioSPINspired 682955, the French National Research Agency (ANR) under grant MEMOS ANR-14-CE26-0021, and a public grant overseen by the ANR as part of the ‘Investissements d’Avenir’ programme (Labex NanoSaclay, reference ANR-10-LABX-0035).

Reviewer information

Nature thanks F. Hoppensteadt, A. Kent and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Author information

Author notes

    • Flavio Abreu Araujo

    Present address: Institute of Condensed Matter and Nanosciences, UC Louvain, Louvain-la-Neuve, Belgium

  1. These authors contributed equally: Miguel Romera, Philippe Talatchian

Affiliations

  1. Unité Mixte de Physique, CNRS, Thales, Université Paris-Sud, Université Paris-Saclay, Palaiseau, France

    • Miguel Romera
    • , Philippe Talatchian
    • , Flavio Abreu Araujo
    • , Vincent Cros
    • , Paolo Bortolotti
    • , Juan Trastoy
    • , Maxence Ernoult
    •  & Julie Grollier
  2. National Institute of Advanced Industrial Science and Technology (AIST), Spintronics Research Center, Tsukuba, Ibaraki, Japan

    • Sumito Tsunegi
    • , Kay Yakushiji
    • , Akio Fukushima
    • , Hitoshi Kubota
    •  & Shinji Yuasa
  3. Centre de Nanosciences et de Nanotechnologies, CNRS, Université Paris-Sud, Université Paris-Saclay, Orsay, France

    • Maxence Ernoult
    • , Damir Vodenicarevic
    • , Tifenn Hirtzlin
    • , Nicolas Locatelli
    •  & Damien Querlioz

Authors

  1. Search for Miguel Romera in:

  2. Search for Philippe Talatchian in:

  3. Search for Sumito Tsunegi in:

  4. Search for Flavio Abreu Araujo in:

  5. Search for Vincent Cros in:

  6. Search for Paolo Bortolotti in:

  7. Search for Juan Trastoy in:

  8. Search for Kay Yakushiji in:

  9. Search for Akio Fukushima in:

  10. Search for Hitoshi Kubota in:

  11. Search for Shinji Yuasa in:

  12. Search for Maxence Ernoult in:

  13. Search for Damir Vodenicarevic in:

  14. Search for Tifenn Hirtzlin in:

  15. Search for Nicolas Locatelli in:

  16. Search for Damien Querlioz in:

  17. Search for Julie Grollier in:

Contributions

The study was designed by J.G. and D.Q. Samples were optimized and fabricated by S.T. and K.Y. The main experiments were performed by M.R. and P.T. Spin diode experiments were performed by P.T. and J.T. Numerical simulations were realized by P.T., M.E., M.R., T.H. and D.V. All authors contributed to analysing the results and writing the paper.

Competing interests

The authors declare no competing interests.

Corresponding authors

Correspondence to Damien Querlioz or Julie Grollier.

Extended data figures and tables

  1. Extended Data Fig. 1 Schematic of the experimental set-up.

    The four coupled vortex nano-oscillators are shown. IRFA and IRFB are the microwave currents injected in the strip line by the two microwave sources. HRF is the resulting microwave field. IDC1–4 are the applied direct currents.

  2. Extended Data Fig. 2 Recognition rates obtained by different neural networks on the formant database.

    a, Recognition rates of different neural networks on the formant database as a function of the number of trained parameters. be, Schematics of the simulated neural networks: b, multi-layer perceptron; c, perceptron; d, RNN; and e, LSTM.

  3. Extended Data Fig. 3 Synchronization detection by the spin diode effect.

    a, Rectified direct voltage measured between oscillator electrodes when the external microwave signal is injected in the strip line above the oscillator and its frequency is swept. Here, the direct current through the oscillator is 5 mA, the magnetic field is 585 mT and the injected microwave power is +1 dBm. b, Oscillator spectrum emission measured during the same frequency sweep as a. c, Proposed differential measurement configuration for CMOS-based detection of synchronization-induced rectified voltages. d, Two-stage CMOS circuit. e, The first stage, composed of two differential amplifiers (green), is followed by a gain stage (blue). VDD, supply voltage; GND, ground. f, Energy consumption of the CMOS circuit for one synchronization detection event, as a function of the amplitude of the generated rectified direct voltages.

  4. Extended Data Table 1 Learning rule
  5. Extended Data Table 2 Comparison of CMOS and spin-torque nano-oscillators for neuromorphic computing

Supplementary information

  1. Supplementary Information

    The document describes the numerical simulations that were performed to investigate the important features that oscillators should possess to classify accurately.

  2. Supplementary Data

    This file contains the vowel formant database. The spoken vowels used in this study are characterized by a set of frequencies called formants, that we obtain from a subset of the Hillenbrand database (https://homepages.wmich.edu/~hillenbr/voweldata.html). We use the first three formants (F1, F2 and F3) sampled at four different times of the duration of the spoken vowel: at the steady state and at 20%, 50%, and 80% of the vowel duration, respectively (12 parameters in total). When one of these 12 parameters could not be measured or irresolvable formants mergers occurred, Hillenbrand et al.[28] put a zero in this parameter in the database. For our study, we have removed the vowel utterances whose corresponding set of formants is not complete. Moreover, we use the same number of speakers for each vowel. The resulting formant database comprising 37 female speakers that we used is given. The file also includes the input frequencies fA and fB corresponding to each vowel and the coefficients of the two linear combinations that were used to obtain these input frequencies.

  3. Supplementary Video

    Training spin-torque nano-oscillators to classify. This short video shows the learning process for the spin-torque nano-oscillator based hardware. Here four coupled oscillators are trained to recognize seven American vowels. Each vowel is characterized by two frequencies, fA and fB, and is represented by a dot. The different vowels have different colors. The spread within each vowel is due to the different pronunciations of the 37 speakers. Each vowel is applied to the oscillator system as the sum of two microwave magnetic fields which can phase-lock the oscillators. The different colors in the background map represent the different experimental oscillator synchronization configurations obtained for different input frequencies fA and fB. Recognition is achieved when each vowel cloud corresponds to a single synchronization configuration. At the beginning of the video, the oscillator network is randomly set, and the recognition rate is closed to zero. At each training step, the dc current though each oscillator is modified through a learning rule which brings the oscillator frequency closer to the desired synchronization configuration associated to the vowels that have been applied. At the end of the video, after the networks has learned, the dots corresponding to each vowel are in majority encompassed in different regions of the background map corresponding to different synchronization configurations, which gives a recognition rate of 89% on the training data.

About this article

Publication history

Received

Accepted

Published

DOI

https://doi.org/10.1038/s41586-018-0632-y

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.