WO2014106352A1

WO2014106352A1 - Unit, device and method for simulating neuronal synapse of biont

Info

Publication number: WO2014106352A1
Application number: PCT/CN2013/070277
Authority: WO
Inventors: 缪向水; 李祎; 钟应鹏; 许磊; 孙华军
Original assignee: 华中科技大学
Priority date: 2013-01-04
Filing date: 2013-01-09
Publication date: 2014-07-10
Also published as: CN103078055A; CN103078055B

Abstract

Disclosed are a unit, device and method for simulating a neuronal synapse of a biont based on a chalcogenide compound. The unit comprises a first electrode layer, a functional material layer and a second electrode layer. The first electrode layer receives a first pulse signal, and the second electrode layer receives a second pulse signal. A device can change the conductance thereof according to an input signal, so as to simulate the change of synapse weighting. When the difference value between the frequency of the first pulse signal and that of the second pulse signal is positive or negative, the change of conductance realizes the simulation of the function of spike frequency-dependent synapse plasticity of the neuronal synapse of the biont; and when the peak value of the signal difference between the first pulse signal and the second pulse signal is positive or negative, the change of the conductance realizes the simulation of the function of spike timing-dependent synapse plasticity of the neuronal synapse of the biont. The present invention can achieve the basic function of the neuronal synapse of the biont on a single inorganic device, and provide basic components forming an artificial neural network, and can achieve the effects of increasing the integration degree and reducing the power consumption.

Description

a simulated, synaptic unit and device

[Technical Field]

The present invention is in the field of microelectronic devices and, more particularly, relates to a unit, apparatus and method for simulating biological synapses.

【Background technique】

In a traditional computer based on the von Neumann architecture, the processor and memory are separate and connected by a bus. Such an architecture has the so-called "Von Neumann bottleneck", which is difficult to adapt to the current era of rapid development of information technology with explosive growth in information.

Compared to von Neumann computers, human brain neural information activities are characterized by massive parallelism, distributed storage and processing, self-organization, self-adaptation, and self-learning. Researchers in the fields of traditional artificial neural networks and neuromorphic engineering have also been working to simulate the basic bioelectrical characteristics of neuronal synapses, such as neuron triggering and synaptic plasticity, using nonlinear circuits, FPGAs, and VLSI. More advanced cognitive functions such as pattern recognition and intelligent control break through the von Neumann architecture. In these methods, it takes dozens of transistors, capacitors, and adders to simulate only one neuron, one synapse, and one learning module. However, the human brain includes up to ~10 ¹¹ neurons and ~10 ¹⁵ synapses, and the connections between neurons and synapses are more chaotic and incomparably complex. This traditional neuromorphic engineering is powerless to simulate the human brain, even the mouse brain.

IBM used the "Blue Gene" supercomputer to simulate a cat's cerebral cortical cognitive function using 147,456 processor-architecture neural networks. If neuron signal processing can be implemented in nanodevices, the chip size and power consumption of the devices required to simulate the entire brain can be achieved within the achievable range.

Construction of the neural network neurons and synapses relates to the design and preparation, which proved to be learning and memory stored in the synapses in the brain and the number of synapses is about ¹⁰⁴ times the number of neurons in a conventional VLSI In the neural circuit constructed by the CMOS method, the synapse component occupies the entire circuit surface. More than 80% of the product, and consumes most of the power consumption of the circuit, so there is an urgent need for a simple structure, small size, low power consumption components that can achieve synaptic function. The publication number is CN101770560A, and the patent application file for the information processing method and apparatus for simulating the biological neuron information processing mechanism mentions that a plurality of transistors based on a CMOS integrated circuit constitute one neuron, and does not involve a nerve having learning ability. Synapse. The publication number is CN1670963A, and the inventor's name is: The patent application file of a flexible triode that resembles a neuron synaptic structure refers to a structure that only simulates a synapse of a neuron, but does not function as a synapse.

[Summary of the Invention]

In view of the deficiencies of the prior art, it is an object of the present invention to provide a unit for simulating a biological synapse, which aims to solve the problem of achieving a synaptic function by using a plurality of components.

The present invention provides a unit for simulating a biological synapse, comprising a first electrode layer, a functional material layer connected to the first electrode layer, and a second electrode layer connected to the functional material layer; The electrode layer is used to simulate presynaptic, the second electrode layer is used to simulate post-synaptic, the material of the functional material layer is a sulfur-based compound, and the conductance of the functional material layer is used to simulate synaptic weight; The first electrode layer applies a first pulse signal to simulate presynaptic stimulation, and a post-synaptic stimulus is simulated by applying a second pulse signal to the second electrode layer.

Further, the first electrode layer is configured to receive an external first pulse signal, and the second electrode layer is configured to receive an external second pulse signal; when the amplitude of the first pulse signal is different from the first When the difference between the amplitudes of the two pulse signals is positive or negative, the conductance of the functional material layer is changed to realize the simulation of the synaptic weight adjustment function of the biological synapse; when the frequency of the first pulse signal When the difference between the frequency of the second pulse signal is positive or negative, the conductance of the functional material layer changes to achieve a simulation of the pulse rate-dependent synaptic plasticity function of the biological synapse; When the peak of the signal difference between the amplitude of a pulse signal and the amplitude of the second pulse signal is positive or negative, the conductance of the functional material layer changes to achieve pulse time-dependent synaptic plasticity of the biological synapse Functional simulation.

Further, the material of the first electrode layer is an inert conductive metal; the second electrode The material of the layer is a lively conductive metal.

Further, the first electrode layer, the functional material layer and the second electrode layer constitute a sandwich laminate structure, a T-type structure, an I-type structure or a pyramid-type structure.

The present invention also provides an apparatus for simulating a biological synapse, comprising a plurality of array-arranged neurosynaptic units and a controller coupled to the synaptic unit, the synaptic unit being the unit described above.

Further, the controller is configured to apply a first pulse signal to the first electrode layer, a second pulse signal to the second electrode layer, and control a magnitude of the first pulse signal and the first The difference between the amplitudes of the two pulse signals is positive or negative, and the difference between the frequency of the first pulse signal and the frequency of the second pulse signal is controlled to be positive or negative, and the first pulse is controlled. The peak of the signal difference between the amplitude of the signal and the amplitude of the second pulse signal is positive or negative.

The present invention also provides a method of simulating a biological synapse, comprising the steps of: applying a first pulse signal on a first electrode layer and applying a second pulse signal on a second electrode layer;

Adjusting the change in conductance of the functional material layer and simulating synapses of the biological synapse by controlling the positive or negative of the difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal Weight adjustment function;

Adjusting the change in conductance of the functional material layer by controlling the difference between the frequency of the first pulse signal and the frequency of the second pulse signal to be positive or negative and simulating a pulse rate dependent protrusion of the biological synapse Touch plasticity function;

Adjusting the change in conductance of the functional material layer and simulating the pulse of the biological synapse by controlling whether the peak of the signal difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal is positive or negative Time depends on synaptic plasticity.

Further, the step of synthesizing the synaptic weight adjustment function of the biological synapse is specifically: controlling the difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal to be positive, Reducing the conductance of the functional material layer, simulating synapses of biological synapses Weight reduction function;

By controlling the difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal to be negative, the conductance of the functional material layer is increased, simulating the synaptic weight of the biological synapse Up function.

Further, the step of synthesizing the synaptic weight adjustment function of the biological synapse further comprises: controlling a magnitude of a positive difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal The value is increased such that the slower the conductance of the functional material layer is reduced, simulating the slower function of the synaptic weight of the biological synapse falling;

By controlling the magnitude of the negative difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal, the conductance of the functional material layer is increased faster, simulating the creature The faster the synaptic weight of the synapse rises.

Further, the pulse rate dependent synaptic plasticity function of the simulated biological synapse comprises:

Controlling the frequency of the first pulse signal to remain unchanged, by controlling the frequency of the second pulse signal to be a set frequency threshold, so that the conductance of the functional material layer is unchanged, simulating the synapse of the biological synapse a function in which the weight remains the same;

By controlling the frequency of the second pulse signal to be greater than the frequency threshold, the conductance of the functional material layer is increased, simulating a function of synaptic weight rise of the biological synapse;

By controlling the frequency of the second pulse signal to be less than the frequency threshold, the conductance of the layer of functional material is reduced, simulating the function of synaptic weight reduction of biological synapses.

Further, the pulse rate dependent synaptic plasticity function of the simulated biological synapse further comprises:

Controlling and increasing the frequency of the second pulse signal, the faster the conductance of the functional material layer is increased, simulating the function that the synaptic weight of the biological synapse rises faster;

The frequency of the second pulse signal is controlled and reduced, and the slower the conductance of the functional material layer is reduced, simulating the slower function of the synaptic weight of the biological synapse. Further, the pulse time-dependent synaptic plasticity functional steps of the simulated biological synapse include:

Controlling a time difference between the first pulse signal and the second pulse signal to be greater than zero and adjusting a shape of the first pulse signal and the second pulse signal such that a magnitude of the first pulse signal and the first The signal difference peak between the amplitudes of the two pulse signals is negative, and the conductance of the functional material layer is increased, simulating the function of synaptic weight increase of the biological synapse;

Controlling a time difference between the first pulse signal and the second pulse signal to be less than zero and adjusting a shape of the first pulse signal and the second pulse signal such that a magnitude of the first pulse signal and the first The peak of the signal difference between the amplitudes of the two pulse signals is positive, and the conductance of the functional material layer is reduced, simulating the function of the synaptic weight of the biological synapses being small.

Further, the pulse time-dependent synaptic plasticity function of the simulated biological synapse comprises:

Controlling a time difference between the first pulse signal and the second pulse signal to be greater than zero and adjusting a shape of the first pulse signal and the second pulse signal such that a magnitude of the first pulse signal and the first The signal difference peak between the amplitudes of the two pulse signals is positive, and the conductance of the functional material layer is reduced, simulating the function of synaptic weight reduction of the biological synapses;

Controlling a time difference between the first pulse signal and the second pulse signal to be less than zero and adjusting a shape of the first pulse signal and the second pulse signal such that a magnitude of the first pulse signal and the first The signal difference peak between the amplitudes of the two pulse signals is negative, and the conductance of the functional material layer is increased, simulating the function of synaptic weight increase of the biological synapses.

Controlling an absolute value of a time difference of the first pulse signal and the second pulse signal to be less than a quarter of a width of the second pulse signal and adjusting shapes of the first pulse signal and the second pulse signal, The peak of the signal difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal is made negative, the conductance of the functional material layer is increased, and the biological synapse is simulated. The function of synaptic weighting;

Controlling an absolute value of a time difference between the first pulse signal and the second pulse signal to be greater than or equal to a quarter of a width of the second pulse signal and adjusting a shape of the first pulse signal and the second pulse signal And causing a signal difference peak between the amplitude of the first pulse signal and the amplitude of the second pulse signal to be positive, and the conductance of the functional material layer is reduced, simulating the synaptic weight of the biological synapse Reduced functionality.

Controlling an absolute value of a time difference of the first pulse signal and the second pulse signal to be less than one-half of a width of the second pulse signal and adjusting shapes of the first pulse signal and the second pulse signal, And a signal difference peak between the amplitude of the first pulse signal and the amplitude of the second pulse signal is greater than a peak value of the first pulse signal, and a conductance of the functional material layer is reduced, simulating a biological nerve The function of synaptic synaptic weight reduction;

Controlling an absolute value of a time difference of the first pulse signal and the second pulse signal to be greater than or equal to one-half of a width of the second pulse signal and adjusting shapes of the first pulse signal and the second pulse signal And causing a signal difference peak between the amplitude of the first pulse signal and the amplitude of the second pulse signal to be less than or equal to a peak value of the first pulse signal, and the conductance of the functional material layer is unchanged, simulating The function of synaptic weights of biological synapses is unchanged.

Compared with the prior art, the present invention can realize the basic functions of biological synapses in a single device, namely synaptic weight adjustment function, pulse rate-dependent synaptic plasticity function and pulse time-dependent synaptic plasticity function; Degree, reduce the beneficial effects of power consumption.

[Description of the Drawings]

1 is a schematic structural view of a device for simulating a biological synapse according to an embodiment of the present invention; and FIG. 2(a) is a schematic structural view of a unit for simulating a biological synapse provided by Embodiment 1 of the present invention;

2(b) is a controller in a device for simulating a biological synapse provided in Embodiment 1 of the present invention. a diagram of the relationship between the voltage pulse signal and the conductance;

Fig. 2 (c) is a time-correlation diagram of the pre- and post-synaptic voltage pulse signals of the odd-symmetric type I STDP controller simulating biosynaptic synapses in the apparatus for simulating biological synapses provided in the embodiment 1 of the present invention;

Fig. 2(d) is a diagram showing the effect of the odd-symmetric type I STDP of the simulated biological synapse provided by the embodiment 1 of the present invention.

Figure 2 (e) is a time-correlation diagram of pre- and post-synaptic voltage pulse signals of an odd-symmetric type II STDP controller simulating biosynaptic synapses in a device for simulating a biological synapse provided in Example 1 of the present invention;

Fig. 2(f) is a diagram showing the effect of the odd-symmetric type II STDP of the simulated biological synapse provided in the first embodiment of the present invention.

Figure 2 (g) is a time-correlation diagram of the pre- and post-synaptic voltage pulse signals of the even-symmetric type I STDP controller simulating biosynaptic synapses in the device for simulating biological synapses provided in Example 1 of the present invention;

Fig. 2(h) is a diagram showing the effect of the even-symmetric type I STDP of the simulated biological synapse provided by the embodiment 1 of the present invention.

Figure 2 (i) is a time-correlation diagram of pre- and post-synaptic voltage pulse signals of an even symmetric Type II STDP controller simulating biosynaptic synapses in a device for simulating a biological synapse provided in Example 1 of the present invention;

Fig. 2(j) is a diagram showing the effect of the even symmetric type II STDP of the simulated biological synapse provided in the first embodiment of the present invention.

Fig. 2 (k) is a diagram showing the SRDP effect of the simulated biological synapse provided in Example 1 of the present invention. Figure 3 (a) is a schematic view showing the structure of a unit for simulating a biological synapse provided in Example 2 of the present invention;

3(b) is a diagram showing a relationship between a voltage pulse signal and a conductance of a controller in a device for simulating a biological synapse according to Embodiment 2 of the present invention; 3(c) is a time-correlation diagram of pre- and post-synaptic voltage pulse signals of an odd-symmetric type I STDP controller simulating a biological synapse in a device for simulating a biological synapse provided in Embodiment 2 of the present invention;

Fig. 3(d) is a diagram showing the effect of the odd-symmetric type I STDP of the simulated biological synapse provided in the second embodiment of the present invention.

Fig. 3(e) is a diagram showing the effect of SRDP simulating a biological synapse provided in Example 2 of the present invention. Figure 4 (a) is a schematic view showing the structure of a unit for simulating a biological synapse provided in Example 3 of the present invention;

4(b) is a diagram showing a relationship between a voltage pulse signal and a conductance of a controller in a device for simulating a biological synapse provided in Embodiment 3 of the present invention;

Figure 4 (c) is a time-correlation diagram of pre- and post-synaptic voltage pulse signals of an odd-symmetric type I STDP controller simulating biosynaptic synapses in a device for simulating biological synapses provided in Example 3 of the present invention;

Fig. 4 (d) is a diagram showing the effect of the odd-symmetric type I STDP of the simulated biological synapse provided in the third embodiment of the present invention.

Fig. 4 (e) is a diagram showing the SRDP effect of the simulated biological synapse provided in Example 3 of the present invention. [specific lung type]

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The present invention provides a simulated biological synaptic device capable of simulating synaptic plasticity regulation of biological synapses to achieve synaptic inhibition and facilitation. It is a two-terminal device, has a simple structure, and the functional material used is a sulfur-based compound material, which has been maturely applied in the integrated circuit industry, is easy to prepare, and has low cost; the device size can be up to nanometer level, low power consumption, and large The possibility of applying to large-scale neural network arrays. The device for simulating biological synapses of the present invention can simulate the basic functions of biological synapses, and specifically includes: (1) synaptic weights can be based on The positive and negative of the input signal are changed; (2) the synaptic weight can be changed according to the time difference of the pulse before and after the synapse, that is, the pulse time-dependent synaptic plasticity STDP function is realized; (3) the synaptic weight can be based on the frequency difference of the pre- and post-synaptic pulses The change, ie the pulse rate dependent synaptic plasticity SRDP function is achieved.

The synapse device is a two-terminal device with a presynaptic end at one end and a postsynaptic end at the other end. It has a continuously variable conductance value that represents synaptic weight, the strength of the connection between presynaptic and postsynaptic neurons. The conductance value changes according to the direction of the current passing through it, the forward current causes its conductance to decrease, and the negative current increases its conductance; but when the current is less than a certain threshold (ΙΟΟμΑ), its conductance does not change. Pulse-time-dependent synaptic plasticity (STDP) is achieved by designing presynaptic and postsynaptic impulse stimulation signals; achieving pulse rate-dependent synaptic plasticity SRDP (spike-dependent dependent) Plasticity, pulse rate dependent synaptic plasticity) function. The specific implementation is illustrated by the embodiments.

The simulated synaptic device of the present invention can simulate some basic functions of biological synapses, and can provide a basic unit for constructing an artificial neural network.

The invention will now be described more fully hereinafter with reference to the accompanying drawings in which, However, the present invention may be embodied in many different forms, and the invention should not be construed as being limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be more thorough and comprehensive, and The skilled person fully conveys the concepts of the present invention.

The device for simulating a biological synapse includes a controller and a simulated biological synaptic device; the simulated biological synaptic device comprises: a first electrode made of an inert conductive metal such as platinum (Pt), titanium tungsten (TiW), and tantalum (Ta), etc.; the second electrode, the material of which is an active conductive metal such as silver (Ag), copper (Cu), etc.; a functional material whose material is a sulfur-based compound such as Ge ₂ Sb ₂ Te ₅ , Sb ₂ Te ₃ , GeTe, BiTe and AglnSbTe, etc. The conductance G adjustment of the device is based on the oxidation reaction at the interface between the metal electrode and the sulfur-based compound, and the generated active metal ions enter the functional material under the action of the electric field, and the applied voltage of different polarities causes the active metal ions to migrate in different directions to realize the device. The increase or decrease in conductance. The controller can generate electrical signals that form presynaptic and post-synaptic impulses.

Biologically, synapses include presynaptic and postsynaptic. In the present invention, the presynaptic is the first One electrode, the second electrode after the synapse. Presynaptic and postsynaptic stimulation can alter synaptic weights W. The synaptic weight W is expressed by the conductance G of the device, ie W=G. The signal applied to the first electrode is a presynaptic stimulus; the signal applied to the second electrode is a post-synaptic stimulus. Δΐ is the time difference between presynaptic and postsynaptic stimuli. When presynaptic stimuli precede presynaptic stimuli, Δΐ>0; when presynaptic stimuli lag behind posterior synaptic stimuli, A Oo AW is the amount of change in the weight of the synapse that is stimulated.

1 is a schematic structural diagram of a device for simulating a biological synapse according to an embodiment of the present invention, the device for simulating a biological synapse includes a plurality of arrays of synaptic units and a controller connected to the synapse unit, and controlling And applying a first pulse signal to the first electrode layer, applying a second pulse signal to the second electrode layer, and controlling a magnitude of the first pulse signal and a magnitude of the second pulse signal a difference between positive or negative, controlling a difference between a frequency of the first pulse signal and a frequency of the second pulse signal to be positive or negative, controlling a magnitude of the first pulse signal and the The peak of the signal difference between the amplitudes of the second pulse signals is positive or negative. The controller can apply an electrical signal to one or more of the cells in the array. The structure of the synaptic unit may be a sandwich laminate structure, a T-type structure, an I-type structure or a pyramid-shaped structure.

In an embodiment of the invention, the unit for simulating a biological synapse comprises a first electrode layer, a functional material layer connected to the first electrode layer, and a second electrode layer connected to the functional material layer; the first electrode layer is used for simulating the protrusion Before the touch, the second electrode layer is used to simulate post-synaptic, the conductance of the functional material layer is used to simulate synaptic weights; by applying a first pulse signal to the first electrode layer to simulate presynaptic stimulation, by giving the The second electrode layer applies a second pulse signal to simulate post-synaptic stimulation; the first electrode layer is for receiving an external first pulse signal, and the second electrode layer is for receiving an external second pulse signal; When the difference between the amplitude of a pulse signal and the amplitude of the second pulse signal is positive or negative, the conductance of the functional material layer is changed to realize the simulation of the synaptic weight adjustment function of the biological synapse; When the difference between the frequency of the first pulse signal and the frequency of the second pulse signal is positive or negative, the conductance of the functional material layer changes to achieve the pulse rate-dependent synaptic plasticity function of the biological synapse. Simulation; peak signal when the difference between the first signal and the second pulse signal when the positive or negative pulse, electrically conductive layer of the functional material change occurs to achieve biological neurite The pulse time of the touch is dependent on the simulation of the synaptic plasticity function.

The material of the first electrode layer may be an inert conductive metal; the material of the second electrode layer may be an active conductive metal; and the material of the functional material layer may be a sulfur-based compound.

In order to facilitate the description of the unit and method for simulating biological synapses provided by the embodiments of the present invention, detailed examples are as follows:

2(a) shows the structure of a unit for simulating a biological synapse provided by Embodiment 1 of the present invention; the structure is a sandwich laminate structure, the first electrode material is platinum (Pt), and the second electrode material is silver ( Ag), the functional material is AgInSbTe.

Referring to Fig. 2(a), the synapse device of the present invention includes a first electrode 101, a second electrode 103, and a functional material 102 between the first electrode 101 and the second electrode 103. The first electrode 101 and the functional material 102, the functional material 102 and the second electrode 103 are in electrical contact, in a sandwich laminate structure. The first electrode material is platinum (PO, the second electrode material is silver (Ag), and the functional material is silver indium germanium (AgInSbTe).

Fig. 2(b) is a diagram showing the synaptic weight adjustment function simulating the realization of a biological synapse in a synapse device according to Example 1. The voltage pulse signal is applied to the first electrode 101, and the second electrode 103 is grounded, and the conductance is the conductance of the synaptic device between the first electrode 101 and the second electrode 103.

Referring to Figure 2(b), the synaptic device has a continuous conductance, i.e., has a continuous synaptic weight value that can vary with the voltage pulse signal. When the pulse signal is positive, the conductance decreases, that is, the synaptic weight decreases; when the pulse signal is negative, the conductance increases, that is, the synaptic weight increases. The stronger the amplitude of the positive voltage pulse signal, the smaller the value of the decrease in conductance, that is, the smaller the synaptic weight, the more the synapse is suppressed; the stronger the negative voltage pulse signal, the larger the value of the conductance increase, that is, the synaptic weight. The bigger the synapse, the easier it is. Implement synaptic weight adjustment.

2(c) and 2(d) are diagrams showing the odd-symmetric type I pulse time-dependent synaptic plasticity STDP function simulating the realization of biological synapses in a synaptic device according to Example 1. Wherein the presynaptic stimulation signal is applied to the first electrode 101, and the post-synaptic stimulation signal is applied to the second electrode 103. The difference in the excitation signal before and after the synapse is the signal difference between the first electrode 101 and the second electrode 103.

Referring to Fig. 2(c), when Δΐ>0, the peak 201 of the difference between the presynaptic excitation and the postsynaptic excitation signal is negative, and the conductance of the synaptic device is increased under the synaptic stimulation before and after the group. , synaptic weight increases, AW > 0. Similarly, when ΔΚ0, the pre-synaptic stimuli and the postsynaptic stimuli are positive, the conductance of the synaptic devices is reduced and the synaptic weight is reduced. AW < 0. Referring to Figure 2 (d), when At>0, AW>0 and decay with the At index, when ΔΚθ, △W<0 and also decay with the Δΐ exponent. Achieving the odd-symmetry of biological synapses I-pulse time-dependent synaptic plasticity STDP function.

Figures 2(e) and 2(f) are diagrams showing the odd-symmetric type II pulse time-dependent synaptic plasticity STDP function that mimics the realization of biological synapses in a synaptic device according to Example 1. The presynaptic stimulation signal is applied to the first electrode 101, and the post-synaptic stimulation signal is applied to the second electrode 103. The difference between the synaptic excitation signals is the signal between the first electrode 101 and the second electrode 103. difference.

Referring to Fig. 2(e), when Δΐ>0, the peak 202 of the difference between the presynaptic excitation and the postsynaptic excitation signal is positive, and the conductance of the synaptic device is reduced under the synaptic stimulation of the group before and after the synapse , synaptic weight decreases, AW < 0. Similarly, when ΔΚ0, the pre-synaptic stimuli and the post-synaptic stimuli signal have a negative peak value. Under the synaptic stimuli of this group, the conductance of the synaptic device increases, and the synaptic weight increases. AW > 0. Refer to Figure 2 (0, when At>0, AW<0 and decay with the At exponent, when ΔΚθ, △W>0 and also decay with the Δΐ exponential. Achieving the symmetry of the biosynaptic type II pulse time-dependent synaptic plasticity STDP function.

Figures 2(g) and 2(h) are diagrams showing the even-symmetric type I pulse time-dependent synaptic plasticity STDP function that mimics the realization of biological synapses in a synaptic device according to Example 1. The presynaptic stimulation signal is applied to the first electrode 101, and the post-synaptic stimulation signal is applied to the second electrode 103. The difference between the synaptic excitation signals is the signal between the first electrode 101 and the second electrode 103. difference.

Referring to FIG. 2(g), when Δΐ is small (the absolute value of Δΐ is less than a quarter of the width of the pulse signal applied to the second electrode layer), the presynaptic excitation and the post-synaptic excitation signal are poor. Peak 203 is negative. Under this group of synaptic stimulation, the conductance of the synaptic device increases, synaptic weight Increase, AW> 0. Similarly, when Δΐ is large (the absolute value of Δΐ is greater than or equal to a quarter of the width of the pulse signal applied to the second electrode layer), the peak of the difference between the presynaptic excitation and the postsynaptic excitation signal is positive. In this group of synaptic stimuli, the conductance of the synaptic device is reduced, and the synaptic weight is reduced, AW < 0. Referring to Fig. 2 (h), when At is small, AW>0, when At is larger, AW<0, AW is normally shifted with Δΐ, and the even symmetric I-pulse time dependence of biological synapses is realized. Synaptic plasticity STDP function.

2(i) and 2(j) are diagrams showing the even symmetric type II pulse time-dependent synaptic plasticity STDP function simulating biosynthesis in a synapse device according to Example 1. The presynaptic stimulation signal is applied to the first electrode 101, and the post-synaptic stimulation signal is applied to the second electrode 103. The difference between the synaptic excitation signals is the signal between the first electrode 101 and the second electrode 103. difference.

Referring to FIG. 2(i), when Δΐ is small (ie, the absolute value of Δΐ is less than one-half of the width of the pulse signal applied to the second electrode layer), the difference between the presynaptic excitation and the post-synaptic excitation signal The peak 204 is larger (ie, greater than the presynaptic stimuli). Under this group of synaptic stimuli, the conductance of the synaptic device is reduced and the synaptic weight is reduced, AW<0. When Δΐ is large (ie, the absolute value of Δΐ is greater than or equal to one-half of the width of the pulse signal applied to the second electrode layer), the peak of the difference between the presynaptic excitation and the post-synaptic excitation signal is small (ie, Less than or equal to the peak of presynaptic stimuli), under the synaptic stimulation of this group, the conductance of the synaptic device is basically unchanged, and the synaptic weight is basically unchanged, 0. Referring to Fig. 2 (j), when At is small, AW<0, when At is larger, AW 0, AW is negatively distributed with Δΐ, and the even symmetric type II pulse time-dependent synapses of biological synapses are realized. Plasticity STDP function.

Figure 2 (k) is a graph showing the pulse rate-dependent synaptic plasticity SRDP function of simulating biosynaptic sensing in a synaptic device according to Example 1. The presynaptic stimulation signal is applied to the first electrode 101, and the postsynaptic stimulation signal is applied to the second electrode 103.

The frequency of the presynaptic excitation signal remains unchanged. When the voltage of the post-synaptic stimulation signal is pulsed at a set value f. (f. According to the specific requirements, set to lHz~50kHz), the conductance of the synaptic device is basically unchanged, the synaptic weight remains basically unchanged, AW O; the frequency of the post-synaptic stimulus signal The rate is greater than f. At the time, the conductance of the synaptic device increases, the synaptic weight increases, Δ ¥>0, and the greater the frequency of the postsynaptic stimulation signal, the greater the increase in synaptic weight, that is, the larger; The frequency of the post-excitation signal is less than f. At the time, the conductance of the synaptic device is reduced, the synaptic weight is decreased, ΔW<0, and the smaller the frequency of the postsynaptic stimulation signal, the greater the decrease in synaptic weight, that is, the smaller the ΔW. The pulse rate of biosynaptic responses is dependent on synaptic plasticity SRDP function.

3(a) shows the structure of a unit simulating a biological synapse provided by Embodiment 2 of the present invention; referring to FIG. 3(a), the synapse device of the present invention includes a first electrode 101, a second electrode 103, and The insulating layer 104 and the functional material 102 between the first electrode 101 and the second electrode 103. The first electrode 101 and the functional material 102, the functional material 102, and the second electrode 103 form an electrical contact in a T-shaped configuration. The first electrode material is 钽( ), the second electrode material is copper (Cu), the functional material is bismuth telluride (GeTe), and the insulating layer material is silicon dioxide (Si0 ₂ ). The controller can generate an electrical signal to the first electrode and the second electrode.

Fig. 3(b) is a diagram showing the synaptic weight adjustment function simulating the realization of a biological synapse in a synapse device according to Embodiment 2. The voltage pulse signal is applied to the first electrode 101, and the second electrode 103 is grounded, and the conductance is the conductance of the synaptic device between the first electrode 101 and the second electrode 103.

Referring to Figure 3(b), the synaptic device is shown to have continuous conductance, i.e., has a continuous synaptic weight value that can vary with the voltage pulse signal. When the pulse signal is positive, the conductance decreases, that is, the weight of the synaptic device decreases; when the pulse signal is negative, the conductance increases, that is, the weight of the synapse device rises. The stronger the amplitude of the positive voltage pulse signal, the smaller the value of the decrease in conductance, that is, the smaller the weight of the synapse device, the more the synapse is suppressed; the stronger the negative voltage pulse signal, the larger the value of the conductance increase, that is, the synapse The greater the weight of the device, the easier the synapse is. Implement synaptic weight adjustment.

Figures 3 (c) and 3 (d) are diagrams showing the odd-symmetric type I pulse time-dependent synaptic plasticity STDP function that mimics the realization of biological synapses in a synaptic device according to Example 2. The presynaptic stimulation signal is applied to the first electrode 101, and the post-synaptic stimulation signal is applied to the second electrode 103. The difference between the synaptic excitation signals is the signal between the first electrode 101 and the second electrode 103. difference. Referring to Fig. 3 (c), when Δΐ>0, the peak 301 of the difference between the presynaptic excitation and the postsynaptic excitation signal is negative, and the conductance of the synaptic device is increased under the synaptic stimulation before and after the group. , synaptic weight increases, AW > 0. Similarly, when ΔΚ0, the pre-synaptic stimuli and the postsynaptic stimuli are positive, the conductance of the synaptic devices is reduced and the synaptic weight is reduced. AW < 0. Referring to FIG. 3(d), when At>0, AW>0 and decay with the At index, when ΔΚθ, ΔW<0 and also decay with the Δΐ index. The odd-symmetric type I pulse-time-dependent synaptic plasticity STDP function of the biological synapse is achieved.

Similarly, Embodiment 2 can also implement the other three kinds of pulse time-dependent synaptic plasticity STDP functions, which will not be described herein.

Fig. 3(e) is a graph showing the pulse rate-dependent synaptic plasticity SRDP function of simulating the realization of biological synapses in a synaptic device according to Example 2. The presynaptic stimulation signal is applied to the first electrode 101, and the postsynaptic stimulation signal is applied to the second electrode 103.

The frequency of the presynaptic excitation signal remains unchanged. When the voltage of the post-synaptic stimulation signal is at a set value f. (f. can be set to lHz~50kHz according to specific requirements), the conductance of the synaptic device is basically unchanged, the synaptic weight remains basically unchanged, and the frequency of the synaptic stimulus signal is greater than f. At the time, the conductance of the synaptic device increases, the synaptic weight increases, Δ ¥>0, and the greater the frequency of the postsynaptic stimulation signal, the greater the increase in synaptic weight, that is, the larger; The frequency of the post-excitation signal is less than f. At the time, the conductance of the synaptic device is reduced, the synaptic weight is decreased, AW < 0, and the smaller the frequency of the postsynaptic stimulation signal, the greater the decrease in synaptic weight, that is, the smaller. The pulse rate of biosynaptic responses is dependent on synaptic plasticity SRDP function.

4(a) shows the structure of a unit for simulating a biological synapse provided in Example 3; the synapse device of FIG. 4(a) includes a first electrode 101, a second electrode 103, a first electrode 101, and a second electrode The insulating layer 104 between the 103 functional material 102, the first electrode 101 and the functional material 102, the functional material 102, the second electrode 103 are in electrical contact, and have an I-type structure. Wherein, the first electrode material is titanium tungsten (TiW), the second electrode material is silver (Ag), and the functional material is germanium (Ge ₂ Sb ₂ Te ₅ ), The insulating layer material is silicon dioxide (Si0 ₂ ). The controller can generate an electrical signal to the first electrode and the second electrode.

Figure 4 (b) According to Example 3, the synaptic weight adjustment function of the biosynaptic is simulated in a synaptic device. Wherein the voltage pulse signal is applied to the first electrode 11, and the second electrode 103 is electrically conducted as the conductance of the first electrode 10 and the second electrode 103 synapse device.

Referring to Figure 4 (b) The synaptic device has a continuous conductance, i.e., has a continuous synaptic weight value and can vary with the voltage pulse signal. When the pulse signal is positive, the conductance is reduced, that is, the weight of the synaptic device is decreased; when the pulse signal is negative, the conductance is increased, that is, the weight of the synaptic device is increased. The stronger the amplitude of the positive voltage pulse signal, the smaller the value of the decrease in conductance, that is, the smaller the weight of the synapse device, the more the synapse is suppressed; the stronger the negative voltage pulse signal, the larger the value of the conductance increase, that is, the synapse The greater the weight of the device, the easier the synapse is. Implement synaptic weight adjustment.

Figures 4(c) and 4(d) are diagrams showing the odd-symmetric type I pulse time-dependent synaptic plasticity STDP function that mimics the realization of biological synapses in a synaptic device according to Example 3. The presynaptic stimulation signal is applied to the first electrode 101, and the post-synaptic stimulation signal is applied to the second electrode 103. The difference between the synaptic excitation signals is the signal between the first electrode 101 and the second electrode 103. difference.

Referring to Fig. 4(c), when Δΐ>0, the peak 401 of the difference between the presynaptic excitation and the postsynaptic stimulation signal is negative, and the conductance of the synaptic device is increased under the synaptic stimulation before and after the group. , synaptic weight increases, AW > 0. Similarly, when ΔΚ0, the pre-synaptic stimuli and the postsynaptic stimuli are positive, the conductance of the synaptic devices is reduced and the synaptic weight is reduced. AW < 0. Referring to Figure 4(d), when At>0, AW>0 and decay with the At index, when ΔΚθ, △W<0 and also decay with the Δΐ exponent. Achieving the odd-symmetry of biological synapses I-pulse time-dependent synaptic plasticity STDP function.

Similarly, Embodiment 3 can also implement the other three pulse time-dependent synaptic plasticity STDP functions, which will not be described in detail herein.

Figure 4 (e) is a graph showing the pulse rate dependent synaptic plasticity SRDP function that mimics the realization of biological synapses in a synaptic device according to Example 3. Where the presynaptic excitation signal is applied to the first On the electrode 101, a post-synaptic stimulation signal is applied to the second electrode 103.

The frequency of the presynaptic excitation signal remains unchanged. When the voltage of the post-synaptic stimulation signal is at a set value f. When nearby, the conductance of the synaptic device is basically unchanged, and the synaptic weight remains basically unchanged, AW O; when the frequency of the post-synaptic stimulation signal is greater than f. At the time, the conductance of the synaptic device increases, the synaptic weight increases, Δ ¥>0, and the greater the frequency of the postsynaptic stimulation signal, the greater the increase in synaptic weight, that is, the larger; The frequency of the post-excitation signal is less than f. At the time, the conductance of the synaptic device is reduced, the synaptic weight is decreased, Δ ¥ < 0, and the smaller the frequency of the postsynaptic stimulation signal, the larger the decrease in synaptic weight, that is, the smaller. The pulse rate at which biosynaptic responses are achieved depends on synaptic plasticity SRDP function.

The unit for simulating biological synapses provided by the embodiments of the present invention can also simulate other various STDP functions of the synaptic pulse-dependent synaptic plasticity function, which will not be described in detail herein.

The device and the method for simulating a biological synapse disclosed by the invention; the simulated synapse device can change its weight state according to the input stimulation signal, and change the weight state according to the time difference of the input excitation signal at both ends, thereby realizing synaptic inhibition and Facilitate, and change the weight state according to the frequency of the input signals at both ends to achieve synaptic suppression and facilitation. The present invention can provide basic components constituting an artificial neural network. Features of the various embodiments described herein can be combined or modified in ways that are not explicitly shown. The present invention has been particularly shown and described with reference to the exemplary embodiments of the present invention, In the premise, the form and details can be changed differently.

Claims

Rights request

1. A unit that simulates biological synapses, characterized in that it includes a first electrode layer, a functional material layer connected to the first electrode layer, and a second electrode layer connected to the functional material layer; The first electrode layer is used to simulate the presynaptic, the second electrode layer is used to simulate the postsynaptic, the material of the functional material layer is a chalcogenide compound, and the conductance of the functional material layer is used to simulate the synaptic weight; Presynaptic stimulation is simulated by applying a first pulse signal to the first electrode layer, and postsynaptic stimulation is simulated by applying a second pulse signal to the second electrode layer.

2. The unit of claim 1, wherein the first electrode layer is used to receive an external first pulse signal, and the second electrode layer is used to receive an external second pulse signal; when the When the difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal is positive or negative, the conductance of the functional material layer changes to realize the synaptic weight adjustment function of the biological synapse. Simulation; When the difference between the frequency of the first pulse signal and the frequency of the second pulse signal is positive or negative, the conductance of the functional material layer changes to realize the pulse rate dependence of the biological synapse. Simulation of synaptic plasticity function; When the signal difference peak value between the first pulse signal and the second pulse signal is positive or negative, the conductance of the functional material layer changes to realize the pulse of the biological synapse Modeling of time-dependent synaptic plasticity functions.

3. The unit according to claim 1, wherein the material of the first electrode layer is an inert conductive metal; and the material of the second electrode layer is an active conductive metal.

4. The unit according to claim 1, wherein the first electrode layer, the functional material layer and the second electrode layer form a sandwich stack structure, a T-shaped structure, an I-shaped structure or a pyramid structure. structure.

5. A device for simulating biological synapses, including a plurality of synapse units arranged in an array and a controller connected to the synapse units, characterized in that, the synapse unit of claim 1 -4 Units described in any of the above items.

6. The device according to claim 5, wherein the controller is used to provide the first Apply a first pulse signal to one electrode layer, apply a second pulse signal to the second electrode layer, and control the difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal to be positive. or negative, control the difference between the frequency of the first pulse signal and the frequency of the second pulse signal to be positive or negative, control the amplitude of the first pulse signal and the amplitude of the second pulse signal The signal difference between the peak values is either positive or negative.

7. A method of simulating biological nerve synapses, characterized by including the following steps: applying a first pulse signal on the first electrode layer, and applying a second pulse signal on the second electrode layer;

By controlling the positive or negative value of the difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal, the change in conductance of the functional material layer is adjusted and the synapse of the biological nerve synapse is simulated. Weight adjustment function;

By controlling the difference between the frequency of the first pulse signal and the frequency of the second pulse signal to be positive or negative, the change in conductance of the functional material layer is adjusted and the pulse rate-dependent synapse of the biological nerve synapse is simulated. Touch plasticity function;

By controlling the peak value of the signal difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal to be positive or negative, the change in conductance of the functional material layer is adjusted and the pulse of the biological nerve synapse is simulated. Time-dependent synaptic plasticity functions.

8. The method of claim 7, wherein the functional step of adjusting the synaptic weight of the simulated biological synapse is specifically:

By controlling the difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal to be positive, the conductance of the functional material layer is reduced, simulating the synaptic weight of the biological synapse. Descending function;

By controlling the difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal to be negative, the conductance of the functional material layer is increased, simulating the synaptic weight of the biological synapse. rising function.

9. The method of claim 8, wherein the simulated biological synapse Synaptic weight adjustment functional steps also include:

By controlling the amplitude enhancement of the positive difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal, the conductance of the functional material layer decreases slower, simulating biological nerves. The synaptic weight of a synapse decreases more slowly;

By controlling the amplitude enhancement of the negative difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal, the conductance of the functional material layer increases faster, simulating biological The synaptic weight of a synapse rises faster the faster it functions.

10. The method of claim 7, wherein the functional step of simulating the pulse rate-dependent synaptic plasticity of biological synapses includes:

By controlling the frequency of the first pulse signal to remain unchanged, and controlling the frequency of the second pulse signal to a set frequency threshold, the conductance of the functional material layer remains unchanged, simulating the synapse of a biological nerve synapse. Functions whose weights remain unchanged;

By controlling the frequency of the second pulse signal to be greater than the frequency threshold, the conductance of the functional material layer is increased, simulating the function of increasing the synaptic weight of the biological synapse;

By controlling the frequency of the second pulse signal to be less than the frequency threshold, the conductance of the functional material layer is reduced, simulating the function of decreasing the synaptic weight of the biological synapse.

11. The method of claim 10, wherein the functional step of simulating the pulse rate-dependent synaptic plasticity of biological synapses further includes:

Controlling the frequency of the second pulse signal and increasing it, the faster the conductance of the functional material layer increases, simulating the function of the synaptic weight of the biological synapse rising faster;

The frequency of the second pulse signal is controlled and reduced. The slower the conductance of the functional material layer decreases, the function of simulating the slower the synaptic weight of the biological synapse decreases.

12. The method of claim 7, wherein the functional step of simulating the pulse time-dependent synaptic plasticity of biological synapses includes:

Control the time difference between the first pulse signal and the second pulse signal to be greater than zero and adjust the shapes of the first pulse signal and the second pulse signal so that the amplitude of the first pulse signal The peak value of the signal difference between the value and the amplitude of the second pulse signal is negative, and the conductance of the functional material layer increases, simulating the function of increasing the synaptic weight of the biological synapse;

Control the time difference between the first pulse signal and the second pulse signal to be less than zero and adjust the shapes of the first pulse signal and the second pulse signal so that the amplitude of the first pulse signal is consistent with the amplitude of the second pulse signal. The signal difference peak value between the amplitudes of the two pulse signals is positive, and the conductance of the functional material layer decreases, simulating the function of a biological synapse with a smaller synaptic weight.

13. The method of claim 7, wherein the functional step of simulating the pulse time-dependent synaptic plasticity of biological synapses includes:

Control the time difference between the first pulse signal and the second pulse signal to be greater than zero and adjust the shapes of the first pulse signal and the second pulse signal so that the amplitude of the first pulse signal is consistent with the amplitude of the second pulse signal. The peak value of the signal difference between the amplitudes of the two pulse signals is positive, and the conductance of the functional material layer decreases, simulating the function of reducing the synaptic weight of the biological synapse;

Control the time difference between the first pulse signal and the second pulse signal to be less than zero and adjust the shapes of the first pulse signal and the second pulse signal so that the amplitude of the first pulse signal is consistent with the amplitude of the second pulse signal. The peak value of the signal difference between the amplitudes of the two pulse signals is negative, and the conductance of the functional material layer increases, simulating the function of increasing the synaptic weight of the biological synapse.

14. The method of claim 7, wherein the functional step of simulating the pulse time-dependent synaptic plasticity of biological synapses includes:

controlling the absolute value of the time difference between the first pulse signal and the second pulse signal to be less than a quarter of the width of the second pulse signal and adjusting the shapes of the first pulse signal and the second pulse signal, The peak value of the signal difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal is negative, and the conductance of the functional material layer increases, simulating the increase in synaptic weight of the biological synapse. Big functions;

Controlling the absolute value of the time difference between the first pulse signal and the second pulse signal to be greater than or equal to a quarter of the width of the second pulse signal and adjusting the shapes of the first pulse signal and the second pulse signal , so that the amplitude of the first pulse signal is equal to the amplitude of the second pulse signal The peak value of the signal difference between the values is positive, and the conductance of the functional material layer decreases, simulating the function of reducing the synaptic weight of biological synapses.

15. The method of claim 7, wherein the functional step of simulating the pulse time-dependent synaptic plasticity of biological synapses includes:

Controlling the absolute value of the time difference between the first pulse signal and the second pulse signal to be less than half the width of the second pulse signal and adjusting the shapes of the first pulse signal and the second pulse signal, The peak value of the signal difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal is greater than the peak value of the first pulse signal, and the conductance of the functional material layer is reduced, simulating biological nerves. function of synaptic weight reduction at synapses;

Control the absolute value of the time difference between the first pulse signal and the second pulse signal to be greater than or equal to half the width of the second pulse signal and adjust the shapes of the first pulse signal and the second pulse signal , so that the peak value of the signal difference between the amplitude of the first pulse signal and the amplitude of the second pulse signal is less than or equal to the peak value of the first pulse signal, the conductance of the functional material layer remains unchanged, simulation The synaptic weight-invariant function of biological synapses is achieved.