IP Library › Granted Patent US 12,387,104
Granted Patent B2
US 12,387,104 · App. 18/138,984 · Granted Aug 12, 2025

Deep learning in bipartite memristive networks

Inventors: Jack D. Kendall (Gainesville, FL); Juan C. Nino (Gainesville, FL); Laura E. Suarez (Bogota, CO)
Assignee: UNIVERSITY OF FLORIDA RESEARCH FOUNDATION, INC.
G06N3/084G06N3/049G06N3/065G06N5/046
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,387,104
App. No.
18/138,984
Granted
Aug 12, 2025
Kind
B2
Abstract

A bipartite memristive network and method of teaching such a network is described herein. In one example case, the memristive network can include a number of nanofibers, wherein each nanofiber comprises a metallic core and a memristive shell. The memristive network can also include a number of electrodes deposited upon the nanofibers. A first set of the number of electrodes can include input electrodes in the memristive network, and a second set of the number of electrodes can include output electrodes in the memristive network. The memristive network can be embodied as a bipartite memristive network and trained according to the method of teaching described herein.

Claims (86)

1. A method to train a memristive network comprising a number of input nodes and a number of output nodes, comprising:

applying an input voltage or current to an input node among the number of input nodes;

grounding an output node among the number of output nodes;

measuring an output current or voltage at the grounded output node;

determining an error delta from the measured output current or voltage and a target current or voltage, where the error delta equals the target current or voltage minus the measured output current or voltage;

applying a threshold voltage or current to the output node after ungrounding the output node, the threshold voltage or current applied for a time period proportional to a magnitude of the error delta;

transforming the error delta into a second error delta;

grounding an input node among the number of input nodes; and

applying the threshold voltage or current to the input node after ungrounding the grounded input node, the threshold voltage or current applied for a second time period proportional to the second error delta.

2. The method of claim 1 , wherein, when the error delta is negative, applying the threshold voltage or current to the output node comprises:

applying a positive threshold voltage or current to the output node for the time period proportional to the error delta; and

applying a negative threshold voltage or current to the output node for the time period proportional to the error delta.

3. The method of claim 1 , wherein, when the error delta is positive, applying the threshold voltage or current to the output node comprises:

reversing a polarity of the input voltage or current applied to the input node;

applying a positive threshold voltage or current to the output node for the time period proportional to the error delta; and

applying a negative threshold voltage or current to the output node for the time period proportional to the error delta.

4. The method of claim 1 , wherein:

the second error delta comprises an error delta voltage or current; and

the method further comprises:

applying the error delta voltage or current to the output node with the grounded input node grounded; and

applying the threshold voltage or current to the input node for the second time period after ungrounding the grounded input node, the second time period being proportional to an absolute value of the error delta voltage or current.

5. The method of claim 4 , wherein, when the input voltage or current applied to the input node was positive, applying the threshold voltage or current to the input node for the second time period after ungrounding the grounded input node comprises:

applying a positive threshold voltage or current to the input node for the second time period proportional to the absolute value of the error delta voltage or current; and

applying a negative threshold voltage or current to the input node for the second time period proportional to the absolute value of the error delta voltage or current.

6. The method of claim 4 , wherein, when the input voltage or current applied to the input node was negative, applying the threshold voltage or current to the input node for the second time period after ungrounding the grounded input node comprises:

reversing a polarity of the error delta voltage or current applied to the output node;

applying a positive threshold voltage or current to the input node for the second time period proportional to the absolute value of the error delta voltage or current; and

applying a negative threshold voltage or current to the input node for the second time period proportional to the absolute value of the error delta voltage.

7. The method of claim 1 , wherein the memristive network further comprises:

a memristive network of memristive nanofibers; and

internal electrodes electrically coupled between the number of input nodes and the number of output nodes by memristive shells of the memristive nanofibers.

8. The method of claim 1 , wherein the method reproduces a backpropagation algorithm for training the memristive network of permissive nanofibers.

9. A memristive network, comprising:

a number of nanofibers, wherein each nanofiber comprises a metallic core and a memristive shell;

a number of electrodes deposited upon the nanofibers, wherein the number of electrodes comprise a number of input nodes and a number of output nodes; and

a training processor configured to:

apply an input voltage or current to an input node among the number of input nodes;

ground an output node among the number of output nodes;

measure an output current or voltage at the grounded output node;

determine an error delta from the measured output current or voltage and a target current or voltage, where the error delta equals the target current or voltage minus the measured output current or voltage;

apply a threshold voltage or current to the output node after ungrounding the grounded output node, the threshold voltage or current applied for a time period proportional to a magnitude of the error delta;

transform the error delta into a second error delta;

ground an input node among the number of input nodes; and

apply the threshold voltage or current to the input node after ungrounding the grounded input node, the threshold voltage or current applied for a second time period proportional to the second error delta.

10. The memristive network according to claim 9 , wherein, when the error delta is negative, the training processor is further configured to:

apply a positive threshold voltage or current to the output node for the time period proportional to the error delta; and

apply a negative threshold voltage or current to the output node for the time period proportional to the error delta.

11. The memristive network according to claim 9 , wherein, when the error delta is positive, the training processor is further configured to:

reverse a polarity of the input voltage or current applied to the input node;

apply a positive threshold voltage or current to the output node for the time period proportional to the error delta; and

apply a negative threshold voltage or current to the output node for the time period proportional to the error delta.

12. The memristive network according to claim 9 , wherein:

the second error delta comprises an error delta voltage or current; and

the training processor is further configured to:

apply the error delta voltage or current to the output node with the grounded input node grounded; and

apply the threshold voltage or current to the input node for the second time period after ungrounding the grounded input node, the second time period being proportional to an absolute value of the error delta voltage or current.

13. The memristive network according to claim 12 , wherein, when the input voltage or current applied to the input node was positive, the training processor is further configured to:

apply a positive threshold voltage or current to the input node for the second time period proportional to the absolute value of the error delta voltage or current; and

apply a negative threshold voltage or current to the input node for the second time period proportional to the absolute value of the error delta voltage or current.

14. The memristive network according to claim 12 , wherein, when the input voltage or current applied to the input node was negative, the training processor is further configured to:

reverse a polarity of the error delta voltage or current applied to the output node;

apply a positive threshold voltage or current to the input node for the second time period proportional to the absolute value of the error delta voltage or current; and

apply a negative threshold voltage or current to the input node for the second time period proportional to the absolute value of the error delta voltage.

15. The memristive network according to claim 9 , wherein the memristive network further comprises:

a memristive network of memristive nanofibers; and

internal electrodes electrically coupled between the number of input nodes and the number of output nodes by memristive shells of the memristive nanofibers.

16. A method to train a memristive network comprising a number of input nodes and a number of output nodes, comprising:

applying an input voltage to an input node among the number of input nodes;

grounding an output node among the number of output nodes;

measuring an output current at the grounded output node;

determining an error delta from the measured output current and a target current to determine an error delta, where the error delta equals the target current or voltage minus the measured output current or voltage;

applying a threshold voltage to the output node after ungrounding the grounded output node, the threshold voltage or current applied for a time period proportional to a magnitude of the error delta;

transforming the error delta into a second error delta;

grounding an input node among the number of input nodes; and

applying the threshold voltage to the input node after ungrounding the grounded input node, the threshold voltage or current applied for a second time period proportional to the second error delta.

17. The method of claim 16 , wherein, when the error delta is negative, applying the threshold voltage to the output node after ungrounding the grounded output node comprises:

applying a positive threshold voltage to the output node for the time period proportional to the error delta; and

applying a negative threshold voltage to the output node for the time period proportional to the error delta.

18. The method of claim 16 , wherein, when the error delta is positive, applying the threshold voltage to the output node after ungrounding the grounded output node comprises:

reversing a polarity of the input voltage applied to the input node;

applying a positive threshold voltage to the output node for the time period proportional to the error delta; and

applying a negative threshold voltage to the output node for the time period proportional to the error delta.

19. The method of claim 16 , wherein the memristive network further comprises:

a memristive network of memristive nanofibers; and

internal electrodes electrically coupled between the number of input nodes and the number of output nodes by memristive shells of the memristive nanofibers.

20. The method of claim 19 , wherein the method reproduces a backpropagation algorithm for training the memristive network of memristive nanofibers.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded May 22, 2023
From: KENDALL, JACK D.; NINO, JUAN C.; SUAREZ, LAURA E.
To: UNIVERSITY OF FLORIDA RESEARCH FOUNDATION, INC.
Reel/Frame 063715/0084 →
Continuity (3)
Continuation 15985212 · May 21, 2018
Provisional Application 62509423 · May 22, 2017
Related Publication 20230297839A1 · Sep 21, 2023
References Cited (78)
US 8930291B1 · Srinivasa et al. · 2015 [cited by applicant]
US 9418331B2 · Gupta et al. · 2016 [cited by applicant]
US 9646243B1 · Gokmen · 2017 [cited by applicant]
US 9691479B1 · Merced Grafals · 2017 [cited by examiner]
US 9847132B1 · Zheng · 2017 [cited by examiner]
US 20040091264A1 · Beshai · 2004 [cited by examiner]
US 20040153426A1 · Nugent · 2004 [cited by applicant]
US 20070291524A1 · Davis · 2007 [cited by examiner]
US 20100081958A1 · She · 2010 [cited by applicant]
US 20110096589A1 · Bratkovski et al. · 2011 [cited by applicant]
US 20120011088A1 · Aparin et al. · 2012 [cited by applicant]
US 20120011090A1 · Tang et al. · 2012 [cited by applicant]
US 20120011093A1 · Aparin et al. · 2012 [cited by applicant]
US 20120194967A1 · Keane et al. · 2012 [cited by applicant]
US 20140151623A1 · Jeon et al. · 2014 [cited by applicant]
US 20140156574A1 · Piekniewski et al. · 2014 [cited by applicant]
US 20140156576A1 · Nugent · 2014 [cited by applicant]
US 20150106316A1 · Birdwell et al. · 2015 [cited by applicant]
US 20150269483A1 · Nishitani · 2015 [cited by examiner]
US 20160004960A1 · Saïghi et al. · 2016 [cited by applicant]
US 20160275395A1 · Amir et al. · 2016 [cited by applicant]
US 20170011290A1 · Taha · 2017 [cited by examiner]
US 20170017877A1 · Kataeva · 2017 [cited by examiner]
US 20170017879A1 · Kataeva et al. · 2017 [cited by applicant]
US 20170098156A1 · Nino et al. · 2017 [cited by applicant]
CN 104779950A · 2015 [cited by applicant]
CN 105160401A · 2015 [cited by applicant]
JP H07282168 · 1995 [cited by applicant]
JP H09185596 · 1997 [cited by applicant]
JP 2013534676 · 2013 [cited by applicant]
KR 1020170019414 · 2017 [cited by applicant]
Nguyen et al., Electrospun nanofibers with a core-shell structure of silicon nanoparticles and carbon nanotubes in carbon for use as lithium-ion battery anodes, Jul. 2014, 8 pages (Year: 2014). [cited by examiner]
CN Office Action for CN 2018800311967 mailed Jan. 19, 2023. [cited by applicant]
PCT Patent Application PCT/US2018/033669 filed on May 21, 2018, International Search Report and Written Opinion mailed on Aug. 6, 2018. [cited by applicant]
PCT Patent Application PCT/US2017/058757 filed on Oct. 27, 2017, International Search Report and Written Opinion mailed on Jan. 18, 2018. [cited by applicant]
PCT Patent Application PCT/US2017/058816 filed on Oct. 27, 2017, International Search Report and Written Opinion mailed on Jan. 26, 2018. [cited by applicant]
European Patent Application EP15810294.7 filed on Jun. 5, 2015, Supplementary European Search Report mailed on Jan. 26, 2018. [cited by applicant]
Deng et al., Deep convex net: A scalable architecture for speech pattern classification. Proceedings of the Annual Conference of the International Speech Communication Association, Interspeech, 2285-2288 (Aug. 2011). [cited by applicant]
Grishin et al., Nb2O5 nanofiber memristor, Applied Physics Letters, 103(5), 053111 (Jul. 2013). [cited by applicant]
Hasler et al., Finding a roadmap to achieve large neuromorphic hardware systems, Frontiers in Neuroscience, 7, 118 (Sep. 2013). [cited by applicant]
Hutchinson et al., Tensor deep stacking networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 35(8), 1944-1957 (Aug. 2013). [cited by applicant]
Izhikevich et al., Large-scale model of mammalian thalamocortical systems. Proceedings of the National Academy of Sciences of the United States of America, 105(9), 3593-8 (Mar. 2008). [cited by applicant]
Jo et al., Nanoscale memristor device as synapse in neuromorphic systems, Nano Letters, 10(4), 1297-1301 (Mar. 2010). [cited by applicant]
Li et al., Fabrication of Titania Nanofibers by Electrospinning. Nano Letters, 3(4), 555-560 (Apr. 2003). [cited by applicant]
Manan Suri et al.; “Bio-Inspired Stochastic Computing Using Binary CBRAM Synapses”; IEEE Transactions on Electron Devices; vol. 60. No. 7; Jun. 4, 2013; pp. 2402-2409. [cited by applicant]
Merolla et al., A digital neurosynaptic core using embedded crossbar memory with 45pJ per spike in 45nm. Custom Integrated Circuits Conference (CICC), 2011 IEEE. IEEE, (Sep. 2011). [cited by applicant]
Moghe et al., Co-axial Electrospinning for Nanofiber Structures: Preparation and Applications, Polymer Reviews, Polymer Reviews 48.2, 353-377 (May 2008). [cited by applicant]
Naderi, M. A., Analysis and classification of EEG signals using spectral analysis and recurrent neural networks, Biomedical Engineering (ICBME), 2010 17th Iranian Conference of. IEEE, (Nov. 2010). [cited by applicant]
Strukov et al., The missing memristor found, Nature, 453(7191), 80-3, (May 2008). [cited by applicant]
Waser et al., Redox-Based Resistive Switching Memories—Nanoionic Mechanisms, Prospects, and Challenges, Advanced Materials, 21(25-26), 2632-2663 (Jul. 2009). [cited by applicant]
Williams, R. S. (n.d.), How We Found the Missing Memristor, Spectrum, IEEE, 45(12), 28-35 (Dec. 2008). [cited by applicant]
Zhi et al., Highly conductive electrospun carbon nanofiber/MnO2 coaxial nano-cables for high energy and power density supercapacitors, Journal of Power Sources, 208, 345-353 (Jun. 2012). [cited by applicant]
Zidan et al., Single-Readout High-Density Memristor Crossbar, Scientific reports 6, 2-10 (Jan. 2016). [cited by applicant]
Astaras, A. “Pulse-Stream Binary Stochastic Hardware for Neural Computation: The Helmoltz Machine.” A Thesis Submitted for the Degree of Doctor of Philosophy, The University of Edinburgh. (pp. 1-40, 150-176), Sep. 13, 2… [cited by applicant]
Ben et al., Neurogrid : A Mixed-Analog-Digital Multichip System for Large-Scale Neural Simulations. Proceedings of the IEEE 102.5, 699-716 May 2014). [cited by applicant]
Kee et al. “Feed-Forward versus Feedback Inhibition in a Basic Olfactory Circuit” In: PLoS Comput Biol 11(10):, Oct. 15, 2015 [online] [retrieved on Dec. 27, 2017 (Dec. 27, 2017)] Retrieved from the Internet: URL: http:… [cited by applicant]
Pantazi et al. “All-Memristive Neuromorphic Computing With Level-Tuned Neurons” In: Nanotechnology 27 (2016), Jul. 26, 2016.[online] [retrieved on Dec. 27, 2017 (Dec. 27, 2017)] Retrieved from the Internet: URL: http://… [cited by applicant]
Burger et al., Hierarchical composition of memristive networks for real-time computing. Nanoscale Architectures (NANOARCH), 2015 IEEE/ACM International Symposium on. IEEE, (Jul. 2015). [cited by applicant]
Chan et al., Spike timing dependent plasticity with memristive synapse in neuromorphic systems, The 2012 International Joint Conference on Neural Networks (IJCNN), 1-6 (Jun. 2012). [cited by applicant]
A. Carnell, “An analysis of the use of Hebbian and Anti-Hebbian spike time dependent plasticity learning functions within the context of recurrent spiking neural networks,” Neurocomputing, vol. 72, No. 4-6, pp. 685-692,… [cited by applicant]
A. E. J. Senador, M. T. Shaw, and P. T. Mather, “Electrospinning of Polymeric Nanofibers: Analysis of Jet Formation,” Mat. Res. Soc. Symp. Proc., 2001, vol. 661. [cited by applicant]
A. Joubert, B. Belhadj, O. Temam, and M. Campus, “Hardware spiking neurons design?: analog or digital??” (2012). [cited by applicant]
K. Szot, M. Rogala, W. Speier, Z. Klusek, a Besmehn, and R. Waser, “TiO2—a prototypical memristive material.,” Nanotechnology, vol. 22, No. 25, p. 254001, Jun. 2011. [cited by applicant]
C. Zamarreño-Ramos, L. A. Camuñas-Mesa, J. A. Pérez-Carrasco, T. Masquelier, T. Serrano-Gotarredona, and B. Linares-Barranco, “On spike-timing-dependent-plasticity, memristive devices, and building a self-learning visua… [cited by applicant]
Chronakis, I. S. (2005). Novel nanocomposites and nanoceramics based on polymer nanofibers using electrospinning process—A review. Journal of Materials Processing Technology, 167(2-3), 283-293. doi:10.1016/j.matprotec.2… [cited by applicant]
J. Matser, “Structured Liquids in Liquid State Machines,” Utrecht University, 2010. [cited by applicant]
J. Sundaramurthy, V. Aravindan, P. S. Kumar, S. Madhavi, and S. Ramakrishna, “Electrospun TiO 2-d Nanofibers as Insertion Anode for Li-Ion Battery Applications,” J. Phys. Chem. C, 2014. [cited by applicant]
Tran, T. “Simulations of Artificial Neural Network With Memristive Devices”; Boise State University; Publication [online]. Dec. 2012 [retrieved Jul. 14, 2018]. [cited by applicant]
W. Maass, T. Natschläger, and H. Markram, “Real-time computing without stable states: a new framework for neural computation based on perturbations.,” Neural Comput., vol. 14, No. 11, pp. 2531-2560, Nov. 2002. [cited by applicant]
Yang, J. J., Pickett, M. D., Li, X., Ohlberg, D. A., Stewart, D. R., & Williams, R. S. (2008). Memristive switching mechanism for metal/oxide/metal nanodevices. Nature Nanotechnology, 3(7), 429-433. doi:10.1038/nnano.20… [cited by applicant]
Z. Biolek, D. Biolek and V. Biolková, “SPICE Model of Memristor with Nonlinear Dopant Drift,” Radioengineering, vol. 18 (2) pp. 210-214. [cited by applicant]
Buzsaki, G. (2006). Rhythms of the Brain. Oxford University Press. [cited by applicant]
Japanese Patent Application JP2016-573557 filed on Jun. 5, 2015, Office Action mailed on Sep. 11, 2018. [cited by applicant]
Nicole Casal Moore, College of Engineering, image processing 1,000 times faster is goal of new $5M contact, U-M Press Release, USA, University of Michigan, Aug. 14, 2013. URL, http://www.eecs.umich.edu/eecs/about/articl… [cited by applicant]
European Search Report and Written Opinion for European U.S. Appl. No. 18/805,503 mailed Mar. 10, 2021. [cited by applicant]
European Examination Report for Application No. 2018273847 mailed Mar. 4, 2022. [cited by applicant]
Alibart, et al., “Pattern classification by memristive crossbar circuits using ex situ and in situ training”, Nature Communications, Jun. 25, 2013. [cited by applicant]
Korean Office Action for Korean Patent Application No. 10-2019-7037028 mailed Dec. 29, 2022. [cited by applicant]