Reconfigurable brain-like neuron circuit based on fast-slow circuit

Abstract

The application discloses a reconfigurable brain-like neuron circuit based on fast-slow circuit, which comprises a fast sub-circuit and a slow sub-circuit, and the fast and slow sub-circuits are coupled to determine a pulse emission mode; the fast sub-circuit comprises a volatile threshold resistance variable device CH1 and a capacitor C1, an input end of the CH1 is connected to a signal source through an input resistor, and an output end of the CH1 is connected to an input end of the slow sub-circuit; one end of the C1 is connected to the input end of the CH1, and the other end is grounded; the slow sub-circuit comprises a volatile threshold resistance variable device CH2, a capacitor C2 and a non-volatile threshold resistance variable device in parallel; the input end of the slow sub-circuit is connected with the fast sub-circuit and used for receiving a pulse current of the fast sub-circuit, the output end of the slow sub-circuit is connected with an output resistor, and a connecting point serves as a pulse output port of the neuron. The neuron circuit constructed by the application has stronger information expression capability, is beneficial to improving the performance of a pulse neural network, and can be used for constructing an intelligent computing system with stronger cognitive capability and computing efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of semiconductor information computing, and in particular to a reconfigurable brain-like neuron circuit based on fast-slow circuit. BACKGROUND

[0002] Based on the traditional architecture of computers, the storage and calculation separation technology is used, and the information storage and calculation processing are executed separately. Although the rapid development of deep learning technology has brought great progress to artificial intelligence, the traditional computing architecture will require a lot of energy consumption. Compared with modern computers, the human brain has stronger cognitive ability and high-order computing ability, and can complete various complex tasks with the advantages of high efficiency, high parallelism and low power consumption. This is because the information processing process of the human brain is different from the traditional computing structure. Based on the neural biological network formed by 10 15 synapses and 10 11 neurons, information is transmitted in the form of pulses, with the characteristics of asynchrony, sparsity, spatiotemporal characteristics and event-driven, and has stronger information processing ability.

[0003] The spiking neural network (SNN) simulating biological neural computing has the characteristics of biological computing, and has the advantages of low power consumption compared with the traditional deep neural network. Neurons, as the computing unit of SNN, play an important role in processing and transmitting information. SNN has various neuron models, such as leaky integrate-and-fire (LIF), Hodgkin-huxey (HH) and Izhikevich models.

[0004] The Chinese patent document with publication number CN115984822A discloses a SNN traffic sign recognition method based on spatial attention and related device, and the neuron encoding layer of the SNN network model includes a soft reset LIF neuron. The Chinese patent document with publication number CN108985447A discloses a hardware pulse neural network system, which adopts neuron types including the Morkov-Pert model, the leaky integrate-and-fire model, the Hodgkin-huxey model or the Izhikevich model.

[0005] At present, the commonly used SNN is based on LIF neurons, which can easily simulate the function of neurons, but the disadvantage is the lack of ability to simulate high-order pulse function, resulting in weak information expression ability of neurons and limiting the computing performance of SNN. In addition, although some HH neuron circuits can realize a variety of neuron firing patterns, they require a large hardware overhead and need to change the circuit elements to complete the switching of the pulse firing behavior of the circuit, which limits the further development of chip integration technology of pulse neurons. SUMMARY

[0006] The application provides a reconfigurable brain-like neuron circuit based on a fast-slow circuit, which simulates and switches different high-order pulse firing behaviors in biological neurons by using high-order circuit dynamic response generated by coupling of the fast-slow circuit and reconfigurable circuit design, and overcomes the technical problems of single function of neuron circuit, non-reconfigurable circuit and difficulty in integration.

[0007] A reconfigurable brain-like neuron circuit based on a fast-slow circuit, comprising a fast sub-circuit and a slow sub-circuit, the fast and slow sub-circuits are coupled to determine the pulse firing mode.

[0008] The fast sub-circuit comprises a volatile threshold variable resistor CH1 and a capacitor C1, the input end of the threshold variable resistor CH1 is connected to the input end of the fast sub-circuit through an input resistor R in The output end of the threshold variable resistor CH1 is connected to the input end of the slow sub-circuit; one end of the capacitor C1 is connected to the input end of the threshold variable resistor CH1, and the other end is grounded; the threshold variable resistor CH1 is used for simulating the voltage-gated ion channel of a biological neuron to generate a pulse current.

[0009] The slow sub-circuit comprises a volatile threshold variable resistor CH2, a capacitor C2 and a non-volatile threshold variable resistor NVTSM connected in parallel; the input end of the slow sub-circuit is connected to the fast sub-circuit for receiving the pulse current of the fast sub-circuit, and the output end of the slow sub-circuit is connected to an output resistor R load The connection point serves as a pulse output port of the neuron; by controlling the external voltage to adjust the resistance of the non-volatile threshold variable resistor NVTSM, the parallel resistance of the two variable resistors dynamically affects the charging and discharging process of the capacitor C2, and then controls the capacitor charging and discharging of the fast sub-circuit and the neuron pulse firing behavior.

[0010] In the application, the signal input is connected to the input port of the fast sub-circuit, the generated input current charges the capacitor C1 of the fast sub-circuit, causing the capacitor potential to rise and the membrane voltage to increase. When the membrane voltage reaches the threshold of the volatile threshold variable resistor, the fast sub-circuit is activated and generates a pulse current, which is finally converted into a pulse voltage output to other neurons through the output resistor.

[0011] The slow sub-circuit input end is connected with the fast sub-circuit, for receiving the pulse current of the fast sub-circuit, and the slow sub-circuit output end is connected with the output resistor, and the connection point is used as the pulse output port of the neuron. After the slow sub-circuit receives the pulse current of the fast sub-circuit, the internal circuit state of the slow sub-circuit changes, which feeds back to affect the dynamic behavior of the fast sub-circuit, and the mutual coupling of the two-order circuits realizes the high-order pulse behavior of the simulated biological neuron. By adjusting the resistance of the non-volatile threshold resistance device in the slow sub-circuit through the control signal, the parallel resistance of the sub-circuit and the potential balance state are changed, which further affects the dynamic behavior of the fast sub-circuit and the pulse firing of the neuron, so as to control the pulse firing mode of the neuron.

[0012] Based on the above design, the firing mode of the neuron circuit is no longer a single pulse firing, but different high-order pulse firing modes are switched through the control signal according to the requirements of the use scene, so as to construct a more intelligent neural computing network as a computing unit.

[0013] Further, in the slow sub-circuit, when the resistance of the non-volatile threshold resistance device NVTSM changes from low resistance to high resistance, the influence on the fast sub-circuit also gradually changes, and the pulse firing behavior of the neuron is fast peak firing, adaptive peak firing, refractory period firing and cluster firing in turn.

[0014] Further, the non-volatile threshold resistance device NVTSM continuously changes the resistance under the regulation of the external voltage (i.e. the control signal), and the resistance state remains stable after the voltage is removed; according to the requirements of the pulse firing behavior, the non-volatile threshold resistance device NVTSM is adjusted to a certain resistance state.

[0015] Further, the volatile threshold resistance devices CH1 and CH2 are the same type of threshold resistance devices, and the resistance change is nonlinear, having a threshold switching characteristic; when the capacitor voltage reaches the threshold value, the resistance of the threshold resistance device suddenly changes from high resistance to low resistance, i.e. the device switch opens quickly; when the capacitor voltage is lower than the threshold value, the resistance of the threshold resistance device suddenly changes from low resistance to high resistance, i.e. the device switch closes quickly.

[0016] Further, the capacitor value of the fast sub-circuit is small, and the response speed is fast, and the capacitor value of the slow sub-circuit is large, and the response speed is slow.

[0017] Further, when the membrane potential of the neuron reaches or exceeds the firing threshold, the resistance of the volatile threshold resistance device in the fast sub-circuit decreases suddenly, the total current in the circuit increases suddenly, the voltage division of the output resistor to the ground increases suddenly, and the pulse voltage is generated.

[0018] Further, the pulse output of the neuron circuit is the voltage division of the output resistor R load load ​One end is connected with the output end of the slow sub-circuit and serves as an output port, and the other end is grounded.

[0019] Further, the difference between the capacitor potential of the fast sub-circuit and the capacitor potential of the slow sub-circuit serves as the membrane potential of the neuron.

[0020] Preferably, the capacitor ratio of the fast sub-circuit and the slow sub-circuit ranges from 0.01 to 1, and the capacitor ratio can affect the current dynamic behavior and the pulse firing behavior of the circuit.

[0021] Further, the non-volatile threshold resistance device NVTSM comprises a gate, a source and a drain, wherein the source and the drain are used to form a parallel connection with the volatile threshold resistance device CH2 and the capacitor C2; the gate serves as a control port and is used to receive an external voltage regulation to change the resistance of the non-volatile threshold resistance device NVTSM, thereby affecting the current balance state of the slow sub-circuit, changing the capacitor potential of the slow sub-circuit, and further affecting the capacitor potential of the fast sub-circuit and the pulse firing behavior of the neuron.

[0022] Further, when the capacitor potential of the slow sub-circuit is lower than the threshold value of the volatile threshold resistance device, the device CH2 will not be opened, the slow sub-circuit receives the output current of the fast sub-circuit, the capacitor potential of the slow sub-circuit gradually rises, and the capacitor potential of the fast sub-circuit also needs to rise to make the membrane potential reach the firing threshold of the neuron. With the resistance of the NVSTM increasing from small to large, the firing behavior of the neuron is in turn a fast peak firing with no obvious change in firing frequency, an adaptive peak firing with a firing frequency decreasing from high to low until stable, and an inter-period firing with a firing frequency decreasing from high to low until a refractory period.

[0023] Further, when the capacitor potential of the slow sub-circuit is higher than the threshold value of the volatile threshold resistance device, the device CH2 will be opened, the slow sub-circuit generates a periodic charging and discharging process, the voltage across the fast sub-circuit also follows a periodic change, the neuron periodically switches between the firing and refractory periods, and the neuron is in a burst firing mode.

[0024] Based on the above circuit design, the resistance change of the non-volatile threshold resistance device will affect the current balance of the capacitor in the slow sub-circuit, make the capacitor of the slow sub-circuit reach different voltage values, and cause the neuron to work in different firing modes. The non-volatile threshold resistance device selects a three-terminal device as the regulation unit, which has the following advantages in the circuit: (1) the resistance regulation range of the three-terminal device is wide, which can effectively regulate the dynamic behavior of the slow sub-circuit; (2) the operating voltage of the non-volatile threshold resistance device matches the working voltage of the neuron; (3) the resistance erasing of the non-volatile threshold resistance device is realized through an independent port, which makes the design of the neuron circuit simpler. By using the characteristics of the non-volatile threshold resistance device, a reconfigurable neuron circuit is constructed.

[0025] Further, the resistance of the non-volatile threshold resistance device is determined according to the requirement of the neuron model.

[0026] Further, the resistance of the non-volatile threshold resistance device affects the leakage current of the capacitor in the slow sub-circuit, and the charging current and the leakage current jointly determine the change of the potential of the capacitor.

[0027] Further, the potential of the capacitor in the fast sub-circuit and the slow sub-circuit directly determines the value of the membrane potential of the neuron, and when the membrane potential reaches the firing threshold, the neuron is activated, and when the membrane potential is lower than the firing threshold, the neuron is at rest.

[0028] Compared with the prior art, the present application has the following beneficial effects:

[0029] The neuron circuit designed in the present application can generate high-order pulse firing behaviors of the simulated biological neuron by utilizing the coupling effect of the fast and slow circuits, which specifically exhibits fast peak firing, adaptive peak firing, refractory period firing and cluster firing. At the same time, the switching of different firing modes is realized by changing the resistance of the non-volatile threshold resistance device through a control signal, which embodies the reconfigurability of the neuron circuit. By using the neuron described in the present application as a basic computing unit to construct a neural computing system simulating the human brain, the technical problems of the existing technology, such as the single function of the neuron circuit and the difficulty of chip integration caused by the non-reconfigurability, are overcome, and a more real, efficient and intelligent computing system is expected to be realized. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 is a structure diagram of a reconfigurable brain-like neuron circuit based on fast and slow circuits constructed by an embodiment of the present application;

[0031] Figure 2 is a reconfigurable pulse firing mode of the neuron circuit constructed by an embodiment of the present application;

[0032] Figure 3 is a dynamic behavior analysis of the neuron circuit constructed by an embodiment of the present application;

[0033] Figure 4 is a neuron circuit constructed by an embodiment of the present application, and is a switching control example of different pulse firing modes;

[0034] Figure 5 is an indirect training SNN constructed by an embodiment of the present application;

[0035] Figure 6 is a direct training SNN constructed by an embodiment of the present application. DETAILED DESCRIPTION

[0036] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be noted that the embodiments described below are intended to facilitate the understanding of the present invention and do not constitute any limitation thereof.

[0037] like Figure 1 As shown, a reconfigurable neuromorphic neuron circuit based on fast and slow circuits is presented. The neuron consists of a fast sub-circuit and a slow sub-circuit. The fast sub-circuit generates neuronal pulses, and its input is connected to an input resistor. Its output is connected to an output resistor via the slow sub-circuit. The voltage across the output resistor to ground serves as the neuron's output node. The slow sub-circuit regulates the neuron's firing mode. In this embodiment, the input resistor is R. in The output resistance is R load The threshold resistive switching device for the fast sub-circuit is CH1, and the capacitor is C1. The threshold resistive switching device for the slow sub-circuit is CH2, and the capacitor is C2. The non-volatile threshold resistive switching device is NVTSM.

[0038] The fast sub-circuit includes CH1 and C1, with the input of CH1 connected to R. in Connected to the signal source, the output of CH1 is connected to the input of the slow sub-circuit. One end of C1 is connected to the input of CH1, and the other end is grounded. The slow sub-circuit consists of CH2, NVTSM, and C2, connected in parallel. Its input is connected to the fast sub-circuit, and its output is connected to R. load .

[0039] Furthermore, the ratio of capacitor C1 to capacitor C2 is 1:10, meaning the dynamic change of capacitance in the slow sub-circuit is slower than that in the fast sub-circuit.

[0040] Furthermore, threshold resistive switching devices are volatile devices, meaning that the resistance change cannot be maintained after the voltage applied across them is removed. The same type of threshold resistive switching device is used in both the fast and slow sub-circuits.

[0041] Furthermore, the non-volatile threshold resistive switching device in the slow sub-circuit is a non-volatile three-terminal device. Its gate is independently controlled as a control port, receiving external voltage regulation to change the device's resistance. The source and drain are connected in parallel with the threshold resistive switching device and the capacitor in the slow sub-circuit. Therefore, the dynamic equilibrium state of the slow sub-circuit can be controlled in a simple way.

[0042] In this embodiment, within the same neuron circuit, switching between different pulse firing modes of the neuron can be achieved simply by adjusting the resistance value of the NVTSM in the slow sub-circuit. Figure 2As shown, when the resistance of NVTSM is less than 2KΩ, the firing pattern of neuron is fast spiking, when the resistance gradually increases to 5KΩ, the firing pattern of neuron changes to adaptive firing, when the resistance continues to increase to 10KΩ, the neuron will appear to have a refractory period firing, and when the resistance increases to 20KΩ, the neuron will appear to have a more high-order cluster firing behavior; the above neuron firing behavior regulation only needs to change the resistance of NVTSM through an independent control port.

[0043] The characteristics of the reconfigurable circuit of the present embodiment are characterized and verified in principle, such as Figure 3 As shown, through the electrical characterization of the dynamic behavior of the neuron circuit, it is illustrated that different NVTSM resistances affect the firing behavior of the neuron by adjusting the C2 potential in the slow sub-circuit. As the resistance of NVTSM changes from small to large, the internal dynamic behavior of the circuit changes, mainly manifested by the gradual increase of the saturation potential of C2, and the pulse mode of the neuron circuit is fast firing, adaptive firing, refractory period firing and cluster firing in turn. When the saturation potential of C2 is small, the slow sub-circuit has little effect on the fast sub-circuit, and the membrane potential of the neuron can always maintain fast charging and discharging, and the neuron firing behavior is fast firing. When the saturation potential of C2 increases, the membrane potential of the neuron is limited by the slow sub-circuit, and the firing frequency gradually decreases, and the neuron firing behavior is adaptive firing. When the saturation potential of C2 continues to increase, after a short firing activity, the membrane potential of the neuron cannot reach the threshold, and the neuron firing behavior is refractory period firing. When the saturation potential of C2 increases to the threshold of CH2, CH2 will have a threshold transition similar to CH1, so that the potential of C2 will have a periodic charging and discharging, resulting in the same periodic change of the membrane potential, and the neuron will have a higher-order cluster firing. Figure 3 The dynamic change behavior of C2 potential under adaptive firing and cluster firing is given in (b).

[0044] Further, the saturation voltage of C2 is determined by the dynamic balance of the current in the slow sub-circuit. When C2 reaches the saturation voltage, the input current of the slow sub-circuit is equal to the leakage current through NVTSM and CH2, and the potential of C2 is maintained stable.

[0045] Further, when the saturation voltage of C2 reaches the threshold of CH2 in the slow sub-circuit, CH2 will suddenly change from high resistance to low resistance, and the leakage current through CH2 will suddenly increase, which is greater than the input current of the slow sub-circuit, so that the potential of C2 will decrease. When the potential of C2 is lower than a certain value, CH2 returns to high resistance, and the input current of the slow sub-circuit is greater than the leakage current, so that the potential of C2 increases to the saturation voltage again; which shows a periodic dynamic potential change.

[0046] Further, the pulse firing mode switching of the neuron circuit is mainly affected by the NVTSM resistance value change, and the input voltage change has a weak effect on the mode switching, mainly affecting the pulse frequency and other firing behaviors in the same firing mode, such as Figure 4 as shown in (a).

[0047] Further, the burst firing behavior is affected by the input signal strength, and as the input voltage value increases, the number of pulses in each cycle of the burst firing behavior of the same neuron increases, such as Figure 4 as shown in (b).

[0048] In this embodiment, the advantages of the burst firing behavior of the neuron in the pulse neural network are verified by software simulation. First, the network performance improvement of the indirectly trained SNN is verified, and the burst neurons are connected according to the VGG16 network structure shown in (a). Figure 5 Compared with the traditional integrated firing (IF) neurons, the burst-based neural network has higher accuracy and lower inference delay, as shown in (b). Figure 5

[0049] Further, the burst neuron also has a significant improvement in the performance of the directly trained SNN. In the self-organizing network based on the STDP rule shown in (a). Figure 6 The burst neuron can accelerate the training of the network and improve the recognition accuracy of the trained neural network. (a) is a structural diagram of a self-organizing network, and (b) is the network performance of different neuron models.

[0050] Further, in this embodiment, the VGG16 neural network simulation adopts a conversion method to convert the trained deep neural network into a pulse neural network. The IF neuron or the burst neuron is used to replace the traditional neuron in the deep neural network, and the performance of the neural network using the IF neuron and the burst neuron is compared. The input data uses the CIFAR-10 dataset.

[0051] Further, in this embodiment, the self-organizing pulse neural network simulation adopts the STDP rule to directly train the weight. The LIF neuron and the burst neuron are used to train and infer the neural network, respectively. The neural network based on the burst neuron has higher efficiency in the training process and higher recognition accuracy.

[0052] Further, the STDP rule is based on the pulse firing time relationship between the presynaptic neuron and the postsynaptic neuron to update the synaptic connection, which has more realistic biological learning characteristics. The burst firing can more effectively complete the training of the neural network due to its more accurate pulse firing behavior. ​

[0053] The neuron circuit design provided by the application utilizes a high-order fast and slow sub-circuit to generate a simulated more realistic biological pulse behavior. The switching threshold device is a volatile threshold resistance device, the mode regulation unit is a non-volatile threshold resistance device, and the resistors, capacitors and the like are mature commercial devices. By regulating the resistance value of the non-volatile threshold resistance device, the switching of the neuron pulse firing mode is realized in a reconfigurable manner. The burst neuron model is constructed by using the cluster firing behavior of the neuron, and in the conversion and direct training of the pulse neural network, the performance of the neural network can be significantly improved, including recognition accuracy, training speed and inference delay.

[0054] The above embodiments have described the technical solutions and beneficial effects of the application in detail. It should be understood that the above description is only a specific embodiment of the application and is not used to limit the application. Any modification, supplement and equivalent replacement made within the principle range of the application should be included in the protection range of the application.

Claims

1. A reconfigurable spiking neuron circuit based on fast-slow circuit, characterized in that, The neuron comprises a fast sub-circuit and a slow sub-circuit, and the fast and slow sub-circuits are coupled to determine a pulse emission mode. The fast sub-circuit comprises a volatile threshold resistance variable device CH1 and a capacitor C1, the input end of the threshold resistance variable device CH1 is connected to the input end of the slow sub-circuit through an input resistor R in The output end of the threshold resistance variable device CH1 is connected to the input end of the slow sub-circuit; one end of the capacitor C1 is connected to the input end of the threshold resistance variable device CH1, and the other end is grounded; the threshold resistance variable device CH1 is used for simulating the voltage-gated ion channel of a biological neuron to generate a pulse current; The slow sub-circuit comprises a volatile threshold resistance variable device CH2, a capacitor C2 and a non-volatile threshold resistance variable device NVTSM connected in parallel; the slow sub-circuit input end is connected with the fast sub-circuit for receiving the pulse current of the fast sub-circuit, and the slow sub-circuit output end is connected with an output resistor R load The connection point serves as a pulse output port of the neuron; by controlling the external voltage to adjust the resistance of the non-volatile threshold resistance variable device NVTSM, the parallel resistance of the two resistance variable devices dynamically affects the charging and discharging process of the capacitor C2, thereby controlling the capacitor charging and discharging of the fast sub-circuit and the neuron pulse firing behavior.

2. The reconfigurable spiking neuron circuit based on fast-slow circuit according to claim 1, wherein, In the slow sub-circuit, when the resistance of the non-volatile threshold resistance device NVTSM changes from low resistance to high resistance, the influence on the fast sub-circuit also gradually changes, and the pulse emission behavior of the neuron is in turn fast peak emission, adaptive peak emission, refractory period emission and cluster emission.

3. The reconfigurable spiking neuron circuit based on fast-slow circuit according to claim 2, wherein, The non-volatile threshold resistance device NVTSM continuously changes in resistance under external voltage regulation, and the resistance state remains stable after the voltage is removed; according to the requirement of the pulse emission behavior, the non-volatile threshold resistance device NVTSM is adjusted to a certain resistance state.

4. The reconfigurable spiking neuron circuit based on fast-slow circuit according to claim 1, wherein, The volatile threshold resistance devices CH1 and CH2 are threshold resistance devices of the same type, and the resistance change thereof is nonlinear and has a threshold switching characteristic; when the capacitor voltage reaches the threshold, the resistance of the threshold resistance device suddenly changes from high resistance to low resistance, that is, the device switch is quickly opened; when the capacitor voltage is lower than the threshold, the resistance of the threshold resistance device suddenly changes from low resistance to high resistance, that is, the device switch is quickly closed.

5. The reconfigurable spiking neuron circuit based on fast-slow circuit according to claim 1, wherein, The pulse output of the neuron circuit is the voltage division of the output resistor R load The output resistor R load One end is connected with the output terminal of the slow sub-circuit and serves as an output port, and the other end is grounded.

6. The reconfigurable spiking neuron circuit based on fast-slow circuit according to claim 1, wherein, The difference between the capacitor potential of the fast sub-circuit and the capacitor potential of the slow sub-circuit is used as the membrane potential of the neuron.

7. The reconfigurable spiking neuron circuit based on fast-slow circuit according to claim 1, wherein, The capacitor ratio of the fast sub-circuit to the slow sub-circuit ranges from 0.01 to 1. 8.The reconfigurable spiking neuron circuit based on fast-slow circuitry according to claim 1, wherein, The non-volatile threshold resistance device NVTSM comprises a gate, a source and a drain, wherein the source and the drain are used to form parallel connection with the volatile threshold resistance device CH2 and the capacitor C2; the gate is used as a control port to receive external voltage regulation to change the resistance of the non-volatile threshold resistance device NVTSM.

9. The reconfigurable spiking neuron circuit based on fast-slow circuit according to claim 8, wherein, When the capacitor potential of the slow sub-circuit is lower than the threshold of the volatile threshold resistance device, the device CH2 is not opened, the slow sub-circuit receives the output current of the fast sub-circuit, the capacitor potential of the slow sub-circuit gradually rises, which causes the capacitor potential of the fast sub-circuit to also need to rise; as the resistance of NVSTM increases from small to large, the emission behavior of the neuron is in turn fast peak emission with no obvious change in emission frequency, adaptive peak emission with emission frequency decreasing from high to low until a stable state, and refractory period emission with emission frequency decreasing from high to low until a refractory period state.

10. The reconfigurable spiking neuron circuit based on fast-slow circuit according to claim 8, wherein, When the capacitor potential of the slow sub-circuit is higher than the threshold of the volatile threshold resistance device, the device CH2 is opened, the slow sub-circuit generates a periodic charging and discharging process, which causes the voltage across the fast sub-circuit to also follow a periodic change, the neuron periodically switches between the emission and refractory period states, and the neuron is in a cluster emission mode.

Images