Nanotechnology neural network methods and systems
원문보기
IPC분류정보
국가/구분
United States(US) Patent
등록
국제특허분류(IPC7판)
G06E-001/00
G06E-003/00
G06F-015/18
G06G-007/00
출원번호
US-0735934
(2003-12-15)
등록번호
US-7426501
(2008-09-16)
발명자
/ 주소
Nugent,Alex
출원인 / 주소
Knowntech, LLC
대리인 / 주소
Lopez,Kermit D.
인용정보
피인용 횟수 :
93인용 특허 :
51
초록▼
A physical neural network is disclosed, which includes a connection network comprising a plurality of molecular conducting connections suspended within a connection gap formed between one or more input electrodes and one or more output electrodes. One or more molecular connections of the molecular c
A physical neural network is disclosed, which includes a connection network comprising a plurality of molecular conducting connections suspended within a connection gap formed between one or more input electrodes and one or more output electrodes. One or more molecular connections of the molecular conducting connections can be strengthened or weakened according to an application of an electric field across said connection gap. Thus, a plurality of physical neurons can be formed from said molecular conducting connections of said connection network. Additionally, a gate can be located adjacent said connection gap and which comes into contact with said connection network. The gate can be connected to logic circuitry which can activate or deactivate individual physical neurons among said plurality of physical neurons.
대표청구항▼
The embodiments of an invention in which an exclusive property or right is claimed are defined as follows: 1. An electromechanical neural network system based on nanotechnology, comprising: an adaptive synaptic element comprising a plurality of nanoconductors suspended and free to move about in a l
The embodiments of an invention in which an exclusive property or right is claimed are defined as follows: 1. An electromechanical neural network system based on nanotechnology, comprising: an adaptive synaptic element comprising a plurality of nanoconductors suspended and free to move about in a liquid dielectric solution located within a connection gap formed between at least one pre-synaptic electrode and at least one post-synaptic electrode, wherein said liquid dielectric solution comprises a mixture of said plurality of nanoconductors and a dielectric solvent, wherein said liquid dielectric solution possesses an electrical conductance that is less than an electrical conductance of said plurality of nanoconductors suspended in said liquid dielectric solution; a plurality of interconnected nanoconnections associated with said adaptive synaptic element, said plurality of interconnected nanoconnections comprising said plurality of nanoconductors in said liquid dielectric solution, said plurality of interconnected nanoconnections electrically connecting said at least one pre-synaptic electrode to said at least one post-synaptic electrode through said liquid dielectric solution and said plurality of nanoconductors disposed within said liquid dielectric solution; and a voltage mechanism for applying an electric field across said connection gap, whereby said electric field induces a dipole in each nanoconductor among said plurality of nanoconductors, thereby creating a dielectrophoretic force attracting said plurality of nanoconductors to said connection gap and aligning said plurality of nanoconductors within said liquid dielectric solution and strengthening or weakening each nanoconnection among said plurality of interconnected nanoconnections according to an application of said electric field across said connection gap; and a feedback circuit electrically connected to and associated with said adaptive synaptic element and said voltage mechanism, said feedback circuit providing at least one feedback signal, wherein said feedback circuit together with said adaptive synaptic element and said voltage mechanism comprise a multi-layer, feed-forward network that includes said liquid dielectric solution, said at least one pre-synaptic electrode and said at least one-post synaptic electrode. 2. The system of claim 1 further comprising: a plurality of synapses associated with said adaptive synaptic element, wherein said plurality of synapses comprises said plurality of interconnected nanoconnections of said adaptive synaptic element and wherein each synapse among said plurality of synapse adapts to a voltage independent of voltage polarization. 3. The system of claim 1 wherein each nanoconductor among said plurality of nanoconductors comprises a carbon nanotube. 4. The system of claim 1 wherein each nanoconductor among said plurality of nanoconductors comprises gold. 5. The system of claim 1 wherein each nanoconductor among said plurality of nanoconductors comprises latex. 6. The system of claim 1 wherein each nanoconductor among said plurality of nanoconductors comprises DNA. 7. The system of claim 1 wherein each nanoconductor among said plurality of nanoconductors comprises silicon. 8. The system of claim 2 wherein said voltage mechanism for applying an electric field across said connection gap applies a voltage across a space occupied by said liquid dielectric solution to configure and arrange said adaptive synaptic element, said voltage comprising a DC voltage or an AC voltage that when applied across said liquid dielectric solution gradually forms said interconnected nanoconnections of said plurality of interconnected nanoconnections in said liquid dielectric solution within said connection gap between said at least one pre-synaptic electrode and said at least one post-synaptic electrode. 9. The system of claim 1 wherein said feedback circuit connected to said adaptive synaptic element provides for a Hebbian synapse modification that permits said adaptive synaptic element to function as a recurrent and highly interconnected network. 10. The system of claim 1 further comprising a learning mechanism connected to said adaptive synaptic element to train said electromechanical neural network system, wherein said learning mechanism trains said electromechanical physical neural network utilizing an STDP (Spike-Timing Dependent-Plasticity) training rule. 11. The system of claim 8 wherein said voltage mechanism generates a weak alternating electric current perpendicular to said plurality of interconnected nanoconnections, which causes at least some of said plurality of interconnected nanoconnections not contributing to a desired output to weaken and eventually dissolve back into said liquid dielectric solution, thereby allowing for an increased flexibility in a continuous training of said electromechanical neural network system utilizing said training mechanism. 12. The system of claim 10 wherein at least some of said interconnected nanoconnections among said plurality of interconnected nanoconnections are weakened by increasing a temperature of said liquid dielectric solution, which causes at least some of said plurality of interconnected nanoconnections not contributing to a desired output to weaken and eventually dissolve back into said liquid dielectric solution, thereby allowing for an increased flexibility in a continuous training of said electromechanical neural network system utilizing said training mechanism. 13. The system of claim 2 further comprising: at least one base neuron in a perpendicular array structure composed of a plurality of neural network layers coupled with said plurality of synapses comprising said plurality of interconnected nanoconnections of said adaptive synaptic element, wherein each synapse among said plurality of synapses comprises a plurality of connections conduits separated by a particular distance wherein each connection conduit among said plurality of connection conduits is a result of said plurality of nanoconductors aligning in the presence of said electric field, wherein said electric field is generated by a temporal and sequential firing of said at least one base neuron. 14. The system of claim 1 wherein said dielectric solvent comprises a volatile liquid and further comprising an air-tight seal for confining said liquid dielectric solution within said connection gap to prevent said volatile liquid from coming into contact with air. 15. The system of claim 2 further comprising a gate located adjacent said connection gap, and insulated from electrical contact by an insulation layer, wherein said gate is connected to logic circuitry which can activate or deactivate at least one synapse among said plurality of synapses utilizing a gate voltage provided by said gate, whereby a resistance between said at least one pre-synaptic electrode and said at least one post-synaptic electrode is modifiable by aligning said plurality of nanoconductors in said liquid dielectric solution when said electric field is applied across said connection gap and comprises an alternating electric field. 16. An electromechanical neural network system based on nanotechnology, comprising: a resistive synaptic element comprising a plurality of nanoconductors suspended and free to move about in a liquid dielectric solution located within a connection gap formed between at least one pre-synaptic electrode and at least one post-synaptic electrode, wherein said liquid dielectric solution comprises a mixture of said plurality of nanoconductors and a dielectric solvent, wherein said resistive synaptic element functions as an impermanent interconnect between maid at least one pre-synaptic electrode and said at least one post-synaptic electrode; a plurality of interconnected nanoconnections associated with said resistive synaptic element, said plurality of interconnected nanoconnections comprising said plurality of nanoconductors in said liquid dielectric solution, said plurality of interconnected nanoconnections electrically connecting said at least one pre-synaptic electrode to said at least one post-synaptic electrode through said liquid dielectric solution and said plurality of nanoconductors disposed within said liquid dielectric solution; a plurality of synapses associated with said resistive synaptic element, wherein said plurality of synapses comprises said plurality of interconnected nanoconnections of said resistive synaptic element and wherein each synapse among said plurality of synapses is independent of voltage polarization; and a voltage mechanism for applying an AC electric field across said connection gap, whereby said AC electric field induces a dipole in each nanoconductor among said plurality of nanoconductors only when said plurality of nanoconductors is located within said liquid dielectric solution, thereby generating a dielectrophoretic force attracting said plurality of nanoconductors to said connection gap and aligning said plurality of nanoconductors within said liquid dielectric solution and strengthening or weakening each nanoconnection among said plurality of interconnected nunoconnections according to an application of said AC electric field across said connection gap so that said electromechanical neural network system adapts itself to the requirements of a given situation regardless of the initial state of said plurality of interconnected nanoconnections; and a feedback circuit connected to and associated with said resistive synaptic element and said voltage mechanism, said feedback circuit providing at least one feedback signal, wherein the longer the amount of time said AC electric field is applied across said connection gap and/or the greater the frequency or amplitude of said AC electric field applied across said connection gap, the more nanoconductors among said plurality of nanoconcluctors align and the stronger said interconnected nanoconnections among said plurality of nanoconnections become. 17. The system of claim 16 further comprising a learning mechanism connected to said resistive synaptic element to train said electromechanical neural network system, wherein said learning mechanism includes said feedback circuit and trains said electromechanical physical neural network utilizing an STDP (Spike-Timing Dependent-Plasticity) training rule. 18. A method of forming an electromechanical neural network system based on nanotechnology, comprising: providing a liquid dielectric solution comprising a mixture of a dielectric solvent and a plurality of nanoconductors, wherein each nanoconductor among said plurality of nanoconductors is suspended and free to move about in said liquid dielectric solution; forming a connection gap between at least one pre-synaptic electrode and said at lease post-synaptic electrode; configuring a connection network to comprise said plurality of nanoconductors suspended and free to move about in said liquid dielectric solution located within said connection gap formed between said at least one pre-synaptic electrode and said at least one post-synaptic electrode, wherein said connection network comprises an impermanent interconnect between said at least one pre-synaptic electrode and said at least one post-synaptic electrode; configuring said connection network to comprises a plurality of interconnected nanoconnections with said connection network, wherein said plurality of interconnected nanoconnections comprise said plurality of nanoconductors in said liquid dielectric solution, said plurality of interconnected nanoconnections electrically connecting said at least one pre-synaptic electrode to said at least one post-synaptic electrode through said liquid dielectric solution and said plurality of nanoconductors disposed within said liquid dielectric solution; providing a plurality of synapses from said connection network, wherein said plurality of synapses comprises said plurality of interconnected nanoconnections of said connection network and wherein each synapse among said plurality of synapses is independent of voltage polarization; applying an electric field across said connection gap, whereby said electric field induces a dipole in each nanoconductor among said plurality of nanoconductors only when said plurality of nanoconductors is located within said liquid dielectric solution, thereby generating a clielectrophoretic force attracting said plurality of nanoconductors to said connection gap and aligning said plurality of nanoconductors within said liquid dielectric solution and strengthening or weakening each nanoconnection among said plurality of interconnected nanoconnections according to an application of said electric field across said connection gap; and providing at least one feedback signal to said connection network from a feedback circuit connected to and associated with said connection network and said plurality of synapses. 19. The method of claim 18 wherein the longer the amount of time said electric field is applied across said connection gap and/or the greater the frequency or amplitude of said electric field applied across said connection gap, the more nanoconductors among said plurality of nanoconductors align and the stronger said interconnected nanoconnections among said plurality of nanoconnections become. 20. The method of claim 18 further comprising training said electromechanical neural network system utilizing an STDP (Spike-Timing Dependent-Plasticity) training rule.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (51)
Francisco Jose Ayala, Adaptive neural learning system.
Works George A. (San Diego CA) Hicks William L. (San Diego CA) Kasbo Richard L. (San Diego CA) Muenchau Ernest E. (San Diego CA) Deiss Stephen R. (Encinitas CA), Array processor.
DeYong Mark R. (Las Cruces NM) Findley Randall L. (Austin TX) Eskridge Thomas C. (Las Cruces NM) Fields Christopher A. (Rockville MD), Asynchronous temporal neural processing element.
Uchikawa Yoshiki,JPX ; Furuhashi Takeshi,JPX ; Yamaguchi Masashi,JPX ; Fujime Yoko,JPX, Autonomic system for updating fuzzy neural network and control system using the fuzzy neural network.
Forouhi Abdul Rahim (San Jose CA) Hamdy Esmat Z. (Fremont CA) Hu Chenming (Alamo CA) McCollum John L. (Saratoga CA), Electrically programmable antifuse.
McHardy John (Westlake Village CA) Townsend Carl W. (Los Angeles CA) Higley Lin R. (Laguna Hills CA) Ludwig Frank A. (Rancho Palos Verdes CA), Electrochemical synapses for artificial neural networks.
Nandakishore Kambhatla ; Dimitri Kanevsky ; Wlodek Wlodzimierz Zadrozny, Method and apparatus for executing neural network applications on a network of embedded devices.
Clark Noel A. (Boulder CO) Douglas Kenneth (Boulder CO) Rothschild Kenneth J. (Newton MA), Method for parallel fabrication of nanometer scale multi-device structures.
Castelaz Patrick F. (Yorba Linda CA) Mills Dwight E. (La Verne CA) Woo Steven C. (Stanford CA) Jmaev Jack I. (La Habra CA) Henrikson Tammy L. (Fullerton CA), Neural engine for emulating a neural network.
Schulz, Robert; Boily, Sabin; Joly, Alain; Van Neste, Andre; Alamdari, Houshang, Varistors based on nanocrystalline powders produced by mechanical grinding.
Meier, Philip; Passot, Jean-Baptiste; Ibarz Gabardos, Borja; Laurent, Patryk; Sinyavskiy, Oleg; O'Connor, Peter; Izhikevich, Eugene, Apparatus and methods for control of robot actions based on corrective user inputs.
Meier, Philip; Passot, Jean-Baptiste; Ibarz Gabardos, Borja; Laurent, Patryk; Sinyavskiy, Oleg; O'Connor, Peter; Izhikevich, Eugene, Apparatus and methods for control of robot actions based on corrective user inputs.
Breitwisch, Matthew J.; Lam, Chung Hon; Modha, Dharmendra S.; Rajendran, Bipin, Area efficient neuromorphic system that connects a FET in a diode configuration, and a variable resistance material to junctions of neuron circuit blocks.
Ludwig, Lester F., Molecular transistor driving of nanoscale actuators from differential amplifier circuits compatible with carbon nanotube sensors and transducers.
Ludwig, Lester F., Nanoelectronic differential amplifiers and related circuits having carbon nanotubes, graphene nanoribbons, or other related materials.
Oetringer, Eugen, Neural network, a device for processing information, a method of operating a neural network, a program element and a computer-readable medium.
Akopyan, Filipp; Arthur, John V.; Manohar, Rajit; Merolla, Paul A.; Modha, Dharmendra S.; Molnar, Alyosha; Risk, III, William P., Neuromorphic event-driven neural computing architecture in a scalable neural network.
Akopyan, Filipp; Arthur, John V.; Manohar, Rajit; Merolla, Paul A.; Modha, Dharmendra S.; Molnar, Alyosha; Risk, III, William P., Neuromorphic event-driven neural computing architecture in a scalable neural network.
Hall, Dawn D.; Lamorey, Mark C. H.; Stamper, Anthony K., Normally closed microelectromechanical switches (MEMS), methods of manufacture and design structures.
Hall, Dawn D.; Lamorey, Mark C. H.; Stamper, Anthony K., Normally closed microelectromechanical switches (MEMS), methods of manufacture and design structures.
Friedman, Daniel J.; Kim, Seongwon; Lam, Chung H.; Modha, Dharmendra S.; Rajendran, Bipin; Tierno, Jose A., Producing spike-timing dependent plasticity in a neuromorphic network utilizing phase change synaptic devices.
Friedman, Daniel J.; Kim, Seongwon; Lam, Chung H.; Modha, Dharmendra S.; Rajendran, Bipin; Tierno, Jose A., Producing spike-timing dependent plasticity in a neuromorphic network utilizing phase change synaptic devices.
Friedman, Daniel J.; Kim, Seongwon; Lam, Chung H.; Modha, Dharmendra S.; Rajendran, Bipin; Tierno, Jose A., Producing spike-timing dependent plasticity in a neuromorphic network utilizing phase change synaptic devices.
Coenen, Olivier, Proportional-integral-derivative controller effecting expansion kernels comprising a plurality of spiking neurons associated with a plurality of receptive fields.
Brezzo, Bernard V.; Chang, Leland; Esser, Steven K.; Friedman, Daniel J.; Liu, Yong; Modha, Dharmendra S.; Montoye, Robert K.; Rajendran, Bipin; Seo, Jae-sun; Tierno, Jose A., Reconfigurable and customizable general-purpose circuits for neural networks.
Brezzo, Bernard V.; Chang, Leland; Esser, Steven K.; Friedman, Daniel J.; Liu, Yong; Modha, Dharmendra S.; Montoye, Robert K.; Rajendran, Bipin; Seo, Jae-sun; Tierno, Jose A., Reconfigurable and customizable general-purpose circuits for neural networks.
Brezzo, Bernard V.; Chang, Leland; Esser, Steven K.; Friedman, Daniel J.; Liu, Yong; Modha, Dharmendra S.; Montoye, Robert K.; Rajendran, Bipin; Seo, Jae-sun; Tierno, Jose A., Reconfigurable and customizable general-purpose circuits for neural networks.
Breitwisch, Matthew Joseph; Cheek, Roger W; Lam, Chung Hon; Modha, Dharmendra Shantilal; Rajendran, Bipin, System for electronic learning synapse with spike-timing dependent plasticity using phase change memory.
Nugent, Alex; Molter, Timothy, Thermodynamic random access memory for neuromorphic computing utilizing AHaH (anti-hebbian and hebbian) and memristor components.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.