CN114668408B - Membrane potential data generation method and system - Google Patents
Membrane potential data generation method and system Download PDFInfo
- Publication number
- CN114668408B CN114668408B CN202210578181.0A CN202210578181A CN114668408B CN 114668408 B CN114668408 B CN 114668408B CN 202210578181 A CN202210578181 A CN 202210578181A CN 114668408 B CN114668408 B CN 114668408B
- Authority
- CN
- China
- Prior art keywords
- membrane potential
- pulse
- time
- neuron
- potential data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000012528 membrane Substances 0.000 title claims abstract description 93
- 238000000034 method Methods 0.000 title claims abstract description 38
- 210000002569 neuron Anatomy 0.000 claims abstract description 100
- 230000000638 stimulation Effects 0.000 claims abstract description 54
- 238000001914 filtration Methods 0.000 claims abstract description 21
- 230000008569 process Effects 0.000 claims abstract description 13
- 230000000284 resting effect Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 230000007704 transition Effects 0.000 claims description 3
- 238000004088 simulation Methods 0.000 abstract description 8
- 238000013528 artificial neural network Methods 0.000 description 23
- 230000007246 mechanism Effects 0.000 description 17
- 230000006870 function Effects 0.000 description 11
- 210000004556 brain Anatomy 0.000 description 10
- 230000000694 effects Effects 0.000 description 10
- 230000036279 refractory period Effects 0.000 description 9
- 241000282414 Homo sapiens Species 0.000 description 6
- 238000005265 energy consumption Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000007177 brain activity Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000002102 hyperpolarization Effects 0.000 description 2
- 150000002500 ions Chemical class 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000002336 repolarization Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000003292 diminished effect Effects 0.000 description 1
- 206010015037 epilepsy Diseases 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000028161 membrane depolarization Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000000926 neurological effect Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
- A61B5/377—Electroencephalography [EEG] using evoked responses
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- Signal Processing (AREA)
- Biophysics (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Psychiatry (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Surgery (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Psychology (AREA)
- Power Engineering (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to a membrane potential data generation method and a membrane potential data generation system. The method comprises the steps of obtaining external stimulation information and a pulse at the previous moment; outputting membrane potential data at the current moment by using a neuron model according to external stimulation information; the process of the neuron model for outputting the membrane potential data at the current moment comprises the following steps: converting the external stimulation information from a time domain to a frequency domain with the first time-to-frequency converter; filtering the converted external stimulation information by using the membrane potential filter; converting the pulse at the previous moment from a time domain to a frequency domain by using the second time-frequency converter; filtering the converted pulse at the last moment by using the pulse filter; and outputting the membrane potential data of the current moment according to the filtered external stimulation information and the filtered pulse of the previous moment. The invention can improve the simulation precision of the membrane potential data.
Description
Technical Field
The invention relates to the technical field of neuron simulation, in particular to a membrane potential data generation method and system.
Background
In recent years, development of brain-like computing, brain science, and deep learning is vigorous, and neurons are receiving close attention as the most basic unit. People hope to research related subjects such as intelligence and consciousness by researching the mechanism of biological neurons and simulating the network structure of the brain. The neuron receives an external stimulation signal, processes the external stimulation signal and transmits the external stimulation signal, and the process of simulating and constructing the neuron is a complex process, which not only can be helpful for understanding the operation mechanism of biological nerve cells, but also is beneficial for developing an artificial neural network algorithm with more biological interpretability.
FIG. 4 shows the mechanism of the biological neuron receiving the external stimulation signal, processing, and transmitting the membrane potential change and pulse generation of the biological neuron in the whole process, and the ordinate in FIG. 4 is the membrane potential/mV. In the absence of any external stimulus, the neuronal membrane potential is at resting potential. When receiving external stimuli, from a biological neurological perspective, the channels in the neuron membrane open and specific charged ions enter and exit the channels, resulting in a change in ion concentration inside and outside the neuron. The changes are shown in figure 4 as depolarization, repolarization and hyperpolarization of the neuronal membrane potential. When the membrane potential exceeds the neuron threshold by an external stimulus (such as a sudden strong stimulus or a continuous moderate stimulus), the neuron generates a pulse at that time, and then repolarization and hyperpolarization of the neuron membrane potential occur, i.e., the membrane potential rapidly drops below the resting potential and slowly rises back to the resting potential after a refractory period (absolute refractory period, relative refractory period) has elapsed. During the refractory period, the neurons experience diminished external stimuli, and any external stimuli have little effect on the membrane potential of the neuron.
The existing neuron model can only represent the activity mechanism of part types of neurons in the brain due to the design angle, and has no universality and universality. In addition, neuron models formed by differential equations, such as LIF models, Hodgkin-Huxley models, and Izhikevich models, contain a large number of hyper-parameters, and require experts to manually adjust the hyper-parameters by experience to simulate neuron-specific activity mechanisms, which is time-consuming and labor-consuming. The time-domain-based constructed neuron generation model, such as an impulse response model, and the point process model, require a large number of parameters, because each parameter corresponds to historical information of membrane potential and impulse. The more parameters, the higher the simulation accuracy. Reducing the number of parameters results in a drastic reduction in the simulation accuracy of the model, meaning that the model is not robust.
In order to facilitate deep understanding of the working mechanism of nerve cells and development of new artificial neural network algorithms and hardware circuits, a new simulation method and system are urgently needed.
Disclosure of Invention
The invention aims to provide a membrane potential data generation method and a membrane potential data generation system, which can improve the simulation precision of membrane potential data.
In order to achieve the purpose, the invention provides the following scheme:
a membrane potential data generating method comprising:
acquiring external stimulation information and a pulse at the previous moment; the external stimulus information includes: white noise at the current moment, and a membrane potential or a pulse input at the current moment;
outputting membrane potential data at the current moment by using a neuron model according to external stimulation information; the neuron model includes: the system comprises a first time-frequency converter, a membrane potential filter, a second time-frequency converter and a pulse filter; the process of the neuron model for outputting the membrane potential data at the current moment comprises the following steps:
converting the external stimulation information from a time domain to a frequency domain with the first time-to-frequency converter;
filtering the converted external stimulation information by using the membrane potential filter;
converting the pulse at the previous moment from a time domain to a frequency domain by using the second time-frequency converter;
filtering the converted pulse at the last moment by using the pulse filter;
and outputting the membrane potential data of the current moment according to the filtered external stimulation information and the filtered pulse of the previous moment.
Optionally, the converting, by using the first time-frequency converter, the external stimulation information from a time domain to a frequency domain specifically includes:
using formulas
Determining the converted external stimulation information;
wherein,
the transformed external stimulus information received for neuron j,
for the external stimulus information received by neuron j,
is a Fourier basis function, t is the current time, and k is the term number of the Fourier basis function to the external stimulus.
Optionally, the filtering the converted external stimulation information by using the membrane potential filter specifically includes:
using formulas
Determining filtered external stimulation information;
wherein,
in order to be the weight, the weight is,
filtered external stimulus information for neuron j.
Optionally, the converting, by using the second time-frequency converter, the pulse at the previous time from the time domain to the frequency domain specifically includes:
using formulas
Determining a pulse at the last moment after conversion;
wherein,
the pulse at the last time instant of neuron j,
is a fourier basis function, t-1 is the last time,
the pulse at the last time after the transition of neuron j,lis the number of terms of the fourier basis function for the pulse.
Optionally, the filtering, by using the pulse filter, the pulse at the last time after the conversion specifically includes:
using formulas
Determining a pulse at a last time after filtering;
wherein,
for the last time pulse filtered by neuron j,
are weights.
Optionally, the outputting, according to the filtered external stimulation information and the filtered pulse at the previous time, the membrane potential data at the current time includes:
using formulas
Outputting the membrane potential data at the current moment;
wherein,
is the membrane potential data of the neuron j at the current moment,
is the resting potential of neuron j.
A membrane potential data generating system comprising:
the data acquisition module is used for acquiring external stimulation information and a pulse at the previous moment; the external stimulus information includes: white noise at the current moment, and a membrane potential or a pulse input at the current moment;
the membrane potential data output module is used for outputting membrane potential data at the current moment by utilizing a neuron model according to external stimulation information; the neuron model includes: the system comprises a first time-frequency converter, a membrane potential filter, a second time-frequency converter and a pulse filter; the process of the neuron model for outputting the membrane potential data at the current moment comprises the following steps:
converting the external stimulation information from a time domain to a frequency domain with the first time-to-frequency converter;
filtering the converted external stimulation information by using the membrane potential filter;
converting the pulse at the previous moment from a time domain to a frequency domain by using the second time-frequency converter;
filtering the converted pulse at the last moment by using the pulse filter;
and outputting the membrane potential data of the current moment according to the filtered external stimulation information and the filtered pulse of the previous moment.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
according to the membrane potential data generation method and system provided by the invention, a mechanism that a membrane potential filter and a pulse filter in a neuron model transmit information through oscillation is adopted, so that the activity mechanism of any neuron can be simulated; the mechanism that the neuron transmits signals in a specific frequency is utilized, the signals are converted from a time domain to a frequency domain, and high simulation precision is kept under a small number of parameters.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a schematic flow chart of a method for generating membrane potential data according to the present invention;
FIG. 2 is a schematic diagram of a neuron model according to the present invention;
FIG. 3 is a schematic diagram of a membrane potential data generating system according to the present invention;
fig. 4 is a schematic diagram of a biological neuron receiving an external stimulus signal.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a membrane potential data generation method and a membrane potential data generation system, which can improve the simulation precision of membrane potential data.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a schematic flow chart of a film potential data generating method provided by the present invention, and as shown in fig. 1, the film potential data generating method provided by the present invention includes:
s101, acquiring external stimulation information and a pulse at the previous moment; the external stimulus information includes: white noise at the current moment, and a membrane potential or a pulse input at the current moment;
wherein the external stimulus information may be determined by the following formula:
wherein,
is a white noise source, and is,
is the membrane potential of the neuron i,
is the pulse of neuron i.
S102, outputting membrane potential data at the current moment by using a neuron model according to external stimulation information; as shown in fig. 2, the neuron model includes: the system comprises a first time-frequency converter, a membrane potential filter, a second time-frequency converter and a pulse filter; the process of the neuron model for outputting the membrane potential data at the current moment comprises the following steps:
s201, converting the external stimulation information from a time domain to a frequency domain by using the first time-frequency converter;
s201 specifically includes:
using formulas
Determining the converted external stimulation information;
wherein,
the transformed external stimulus information received for neuron j,
for the external stimulus information received by neuron j,
is a Fourier basis function, t is the current time, and k is the term number of the Fourier basis function to the external stimulus.
S202, filtering the converted external stimulation information by using the membrane potential filter;
s202 specifically includes:
using formulas
Determining filtered external stimulation information; the membrane potential filter is a weighted summation operation aimed at emphasizing frequencies that are significant, while reducing the effect of unwanted frequencies. From a neurobiological point of view, most of the time, information is transferred between neurons by means of low frequency, which is lower in energy consumption and more energy-saving. The membrane potential filter takes advantage of this mechanism and gives large weight to specific frequencies, thereby emphasizing the effect of these frequencies.
Wherein,
the weight is obtained through an optimization algorithm containing L1 regularization, and the sparsity is realized;
filtered external stimulus information for neuron j.
S203, converting the pulse at the previous moment from a time domain to a frequency domain by using the second time-frequency converter;
s203 specifically includes:
using formulas
Determining after conversionA pulse at a previous time;
wherein,
the pulse at the last time instant of neuron j,
is a fourier basis function, t-1 is the last time,
the pulse at the last time after the transition of neuron j,lis the number of terms of the fourier basis function for the pulse. Theoretically, the number of terms of the Fourier basis function for external stimulation is infinite, the number of terms is determined according to the actual situation of the external stimulation or pulse, and the number of terms of the Fourier basis functions corresponding to the external stimulation and the pulse are different.
S204, filtering the converted pulse at the last moment by using the pulse filter;
the method specifically comprises the following steps:
using formulas
Determining a pulse at a last time after filtering; the pulse filter is a weighted summation operation aimed at emphasizing frequencies that are significant, while reducing the effect of unwanted frequencies. From a neurobiological point of view, most of the time, information is transferred between neurons by means of low frequency, which is lower in energy consumption and more energy-saving. The impulse filter uses this mechanism to give large weight to specific frequencies, thereby emphasizing the effect of these frequencies.
Wherein,
for the last time pulse filtered by neuron j,
for weighting, optimization calculation by regularization with L1The method is used for obtaining the product with sparseness.
And S205, outputting the membrane potential data of the current moment according to the filtered external stimulation information and the filtered pulse of the previous moment. The number of pulses of a neuron is related to the refractory period. If the number of pulses of a neuron increases over a period of time, the refractory period is correspondingly extended, i.e., the neuron is less likely to fire pulses. The refractory period of a neuron can explain pulse patterns with adaptive mechanisms, so the refractory period mechanism is indispensable for constructing a generic neuron model. Pulse of last moment
Is incorporated into the model for revealing the refractory period mechanism of neurons.
S205 specifically includes:
using formulas
Outputting the membrane potential data at the current moment;
wherein,
is the membrane potential data of the neuron j at the current moment,
is the resting potential of neuron j.
Fig. 3 is a schematic structural diagram of a film potential data generating system provided by the present invention, and as shown in fig. 3, the film potential data generating system provided by the present invention includes:
a data acquisition module 301, configured to acquire external stimulation information and a pulse at a previous time; the external stimulus information includes: white noise at the current moment, and a membrane potential or a pulse input at the current moment;
a membrane potential data output module 302, configured to output membrane potential data at a current moment according to external stimulation information by using a neuron model; the neuron model includes: the system comprises a first time-frequency converter, a membrane potential filter, a second time-frequency converter and a pulse filter; the process of the neuron model for outputting the membrane potential data at the current moment comprises the following steps:
converting the external stimulation information from a time domain to a frequency domain with the first time-to-frequency converter;
filtering the converted external stimulation information by using the membrane potential filter;
converting the pulse at the previous moment from a time domain to a frequency domain by using the second time-frequency converter;
filtering the converted pulse at the last moment by using the pulse filter;
and outputting the membrane potential data of the current moment according to the filtered external stimulation information and the filtered pulse of the previous moment.
The artificial neural network model is constructed by adopting the method, and the constructed artificial neural network model is used for vehicle identification, face identification and language processing, so that the accuracy of the identification of the artificial neural network model is improved, and the calculated amount is reduced.
The impulse neural network is used as a third generation artificial neural network and is composed of biological neuron models. Compared with the current neural network, the impulse neural network has great advantages in complex data processing and energy consumption convenience. Specifically, thanks to the characteristics of the biological neuron model, the impulse neural network can process data containing a time domain and a space domain simultaneously, and meanwhile, information (with values of 0 and 1) is transmitted through impulses in the impulse neural network, so that compared with the traditional artificial neural network which transmits information by means of floating point numbers, the impulse neural network has the advantages that the required calculation amount is lower, and a large amount of energy consumption is saved.
As a third-generation artificial neural network, the impulse neural network has prominent expressions in various fields in recent years, such as object recognition, face recognition, natural language processing, brain-computer interface, and the like. Specifically, in the field of object recognition, the pulse neural network Spike-based BP SNN formed based on the LIF model obtains similar precision to that of the traditional artificial neural networks VGG and ResNet on MNIST, CIFAR-10 and SVHN data sets, and meanwhile, the calculation amount is lower.
The actual role of the biological neuron model has three aspects. (1) From the perspective of biological neurology and computational neurology, scientists understand the operation mechanism of biological neurons, the interaction mechanism among biological neurons, and the like through a biological neurology model, and also help to correct the defects of the existing biological neuron model by observing the mechanisms. (2) From the perspective of brain science and brain-like computing, in order to explore the mysteries of the brain and understand the relationship between the activities in the brain and human perception, movement and decision, electroencephalogram information such as brain-computer interfaces and the like is used for helping human beings, such as epilepsy treatment and the control of mechanical arms by disabled people. At present, a human brain map is made during key work, namely, each neuron in the human brain and the connection between the neurons are completely simulated through a computer, and the brain activity simulated by the computer is consistent with the human brain activity when any external stimulus is hoped to be applied. The neurons used in the human brain atlas are biological neuron models. (3) From the perspective of artificial neural networks, which are highly simplified models of biological neurons, are depopulated to computational neurology, with the neuron model being a simple weight multiplied by the input (W X), and collocated with the activation function. Originally limited by computer computing power in the last century, intact biological neurons were not used. At present, the artificial neural network has shown strong capability, and the application of the artificial neural network in various fields has excellent performance. However, the artificial neural network also has limitations, such as huge energy consumption, no inference capability, and the like. With the improvement of computer computing power, it is a great trend to replace the current highly simplified neurons with biological neurons in the artificial neural network, and it is hoped that the artificial neural network can save energy and obtain human-like reasoning ability.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (7)
1. A method of generating membrane potential data, comprising:
acquiring external stimulation information and a pulse at the previous moment; the external stimulus information includes: white noise at the current moment, and a membrane potential or a pulse input at the current moment;
outputting membrane potential data at the current moment by using a neuron model according to external stimulation information; the neuron model includes: the system comprises a first time-frequency converter, a membrane potential filter, a second time-frequency converter and a pulse filter; the process of the neuron model for outputting the membrane potential data at the current moment comprises the following steps:
converting the external stimulation information from a time domain to a frequency domain with the first time-to-frequency converter;
filtering the converted external stimulation information by using the membrane potential filter;
converting the pulse at the previous moment from a time domain to a frequency domain by using the second time-frequency converter;
filtering the converted pulse at the last moment by using the pulse filter;
and outputting the membrane potential data of the current moment according to the filtered external stimulation information and the filtered pulse of the previous moment.
2. The method for generating membrane potential data according to claim 1, wherein the converting the external stimulation information from a time domain to a frequency domain by using the first time-frequency converter specifically comprises:
using formulas
Determining the converted external stimulation information;
wherein,
the transformed external stimulus information received for neuron j,
for the external stimulus information received by neuron j,
is a Fourier basis function, t is the current time, and k is the term number of the Fourier basis function to the external stimulus.
3. The method for generating membrane potential data according to claim 2, wherein the filtering the converted external stimulation information by using the membrane potential filter specifically comprises:
using formulas
Determining filtered external stimulation information;
wherein,
in order to be the weight of the weight,
filtered external stimulus information for neuron j.
4. The method for generating membrane potential data according to claim 3, wherein the converting the pulse at the previous time from the time domain to the frequency domain by using the second time-frequency converter specifically comprises:
using formulas
Determining a pulse at the last moment after conversion;
wherein,
the pulse at the last time instant of neuron j,
is a fourier basis function, t-1 is the last time,
the pulse at the last time after the transition of neuron j,lis the number of terms of the fourier basis function for the pulse.
5. The method for generating membrane potential data according to claim 4, wherein the filtering the pulse at the last time after the conversion by using the pulse filter specifically comprises:
using formulas
Determining a pulse at a last time after filtering;
wherein,
for the last time pulse filtered by neuron j,
are weights.
6. The method for generating membrane potential data according to claim 5, wherein outputting membrane potential data at a current time according to the filtered external stimulation information and the filtered pulse at a previous time specifically comprises:
using formulas
Outputting the membrane potential data at the current moment;
wherein,
is the membrane potential data of the neuron j at the current moment,
is the resting potential of neuron j.
7. A membrane potential data generating system, comprising:
the data acquisition module is used for acquiring external stimulation information and a pulse at the previous moment; the external stimulus information includes: white noise at the current moment, and a membrane potential or a pulse input at the current moment;
the membrane potential data output module is used for outputting membrane potential data at the current moment by utilizing a neuron model according to external stimulation information; the neuron model includes: the system comprises a first time-frequency converter, a membrane potential filter, a second time-frequency converter and a pulse filter; the process of the neuron model for outputting the membrane potential data at the current moment comprises the following steps:
converting the external stimulation information from a time domain to a frequency domain with the first time-to-frequency converter;
filtering the converted external stimulation information by using the membrane potential filter;
converting the pulse at the previous moment from a time domain to a frequency domain by using the second time-frequency converter;
filtering the converted pulse at the last moment by using the pulse filter;
and outputting the membrane potential data of the current moment according to the filtered external stimulation information and the filtered pulse of the previous moment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210578181.0A CN114668408B (en) | 2022-05-26 | 2022-05-26 | Membrane potential data generation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210578181.0A CN114668408B (en) | 2022-05-26 | 2022-05-26 | Membrane potential data generation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114668408A CN114668408A (en) | 2022-06-28 |
| CN114668408B true CN114668408B (en) | 2022-08-02 |
Family
ID=82080955
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210578181.0A Active CN114668408B (en) | 2022-05-26 | 2022-05-26 | Membrane potential data generation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114668408B (en) |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004048513A2 (en) * | 2002-05-03 | 2004-06-10 | University Of Southern California | Artificial neural systems with dynamic synapses |
| JP2010123072A (en) * | 2008-11-21 | 2010-06-03 | Nagoya Institute Of Technology | Back propagation learning method for pulse neuron model |
| US10748058B2 (en) * | 2015-12-28 | 2020-08-18 | International Business Machines Corporation | LUT based neuron membrane potential update scheme in STDP neuromorphic systems |
| WO2022073946A1 (en) * | 2020-10-05 | 2022-04-14 | Innatera Nanosystems B.V. | Adaptation of snns through transient synchrony |
| CN112215912B (en) * | 2020-10-13 | 2021-06-22 | 中国科学院自动化研究所 | Saliency map generation system, method and device based on dynamic vision sensor |
| CN113240102B (en) * | 2021-05-24 | 2023-11-10 | 北京灵汐科技有限公司 | Membrane potential updating method of neuron, brain-like neuron device and processing core |
-
2022
- 2022-05-26 CN CN202210578181.0A patent/CN114668408B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN114668408A (en) | 2022-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dominguez-Morales et al. | Deep spiking neural network model for time-variant signals classification: a real-time speech recognition approach | |
| CN112667080A (en) | Electroencephalogram signal unmanned platform intelligent control method based on deep convolution countermeasure network | |
| CN115769228A (en) | Automatic neural network structure constructed by Bayesian graph exploration | |
| CN108319928B (en) | Deep learning method and system based on multi-target particle swarm optimization algorithm | |
| Perrinet et al. | Coding static natural images using spiking event times: do neurons cooperate? | |
| CN111858989A (en) | Image classification method of pulse convolution neural network based on attention mechanism | |
| Wang et al. | Evolving local plasticity rules for synergistic learning in echo state networks | |
| CN112043473B (en) | Parallel nested and autonomous preferred classifier for brain-myoelectricity fusion perception of intelligent artificial limb | |
| Marmarelis et al. | Nonlinear modeling of dynamic interactions within neuronal ensembles using principal dynamic modes | |
| Shagrir | The brain as an input–output model of the world | |
| Nascimento Jr | Artificial neural networks in control and optimization | |
| CN115346676A (en) | Movement function reconstruction dynamic model construction method based on cortical muscle network | |
| CN114668408B (en) | Membrane potential data generation method and system | |
| Larsen et al. | Adaptive regularization in neural network modeling | |
| Löhr et al. | Complex neuron dynamics on the IBM TrueNorth neurosynaptic system | |
| CN108304913A (en) | A method of realizing convolution of function using spiking neuron array | |
| Hasan et al. | On mathematical analysis of pavlovian conditioning learning process using artificial neural network model | |
| Ansari et al. | Adaptive neurofuzzy system for tuberculosis | |
| CN116236209A (en) | Method for recognizing motor imagery electroencephalogram characteristics of dynamics change under single-side upper limb motion state | |
| CN116369945A (en) | Electroencephalogram cognitive recognition method based on 4D pulse neural network | |
| Jankovic | A new simple/spl infin/OH neuron model as a biologically plausible principal component analyzer | |
| Bisht | Design and Development of Mathematical Models for Computational Neuroscience | |
| Li et al. | Multi-class Motor Imagery Classification Method Using One-dimensional Convolutional Neural Networks | |
| Gangopadhyay et al. | A coupled network of growth transform neurons for spike-encoded auditory feature extraction | |
| Joghataie | Extending concepts of mapping of human brain to artificial intelligence and neural networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |