CN106875005A - Adaptive threshold neuronal messages processing method and system - Google Patents
Adaptive threshold neuronal messages processing method and system Download PDFInfo
- Publication number
- CN106875005A CN106875005A CN201710042067.5A CN201710042067A CN106875005A CN 106875005 A CN106875005 A CN 106875005A CN 201710042067 A CN201710042067 A CN 201710042067A CN 106875005 A CN106875005 A CN 106875005A
- Authority
- CN
- China
- Prior art keywords
- current
- threshold
- neuron
- adaptive threshold
- information
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/061—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using biological neurons, e.g. biological neurons connected to an integrated circuit
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Neurology (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Feedback Control In General (AREA)
Abstract
The present invention relates to a kind of adaptive threshold neuronal messages processing method and system, methods described includes：Receiving front-end spiking neuron output information；Read current PRF neuronal messages；According to the front pulse neuron output information and the current PRF neuronal messages, current PRF neuron output information is calculated；Current adaptive threshold variable and threshold potential are read, current adaptive threshold is calculated；When the current PRF neuron output information is more than or equal to the adaptive threshold, export the current PRF neuron output information, and according to the first adaptive threshold updates model modification current adaptive threshold variable, the current PRF neuron output information, and the current adaptive threshold variable according to the second adaptive threshold updates model modification are not exported then otherwise.The present invention with the granting frequency of each neuron in efficient balance whole network, can improve the information processing capability of impulsive neural networks.
Description
Technical field
The present invention relates to artificial neural network technology field, more particularly to adaptive threshold neuronal messages processing method
And system.
Background technology
The artificial neural network research overwhelming majority of today is still in von neumann machine software and high-performance of arranging in pairs or groups
Realized in GPGPU (General Purpose Graphic Processing Units general graphicals processing unit) platform,
The hardware spending of whole process, energy consumption and information processing rate all allow of no optimist.Therefore, neuromorphic calculating field is fast in recent years
Hail exhibition, i.e., using hardware circuit direct construction neutral net so as to simulate the function of brain, it is intended to realize large-scale parallel, low
Energy consumption, the calculating platform of sustainable complex patterns study.
However, in traditional neuromorphic system, how each spiking neuron in the whole neutral net of efficient balance
Granting frequency so that each spiking neuron can play a role in processing information, be in traditional neural network urgently
The problem of solution.
The content of the invention
Based on this, it is necessary to for how the granting frequency of each spiking neuron in the whole neutral net of efficient balance
Problem, there is provided a kind of adaptive threshold neuronal messages processing method and system, methods described includes：
Receiving front-end spiking neuron output information；
Read current PRF neuronal messages；
According to the front pulse neuron output information and the current PRF neuronal messages, current PRF god is calculated
Through first output information；
Current adaptive threshold variable and threshold potential are read, and according to the current adaptive threshold variable and the threshold
Value current potential, calculates current adaptive threshold；
Whether the current PRF neuron output information is judged more than or equal to the adaptive threshold, if so, then defeated
Go out the current PRF neuron output information, and the current adaptive thresholding according to the first adaptive threshold updates model modification
Value variable, if it is not,
Do not export the current PRF neuron output information then, and model modification institute is updated according to the second adaptive threshold
State current adaptive threshold variable.
Wherein in one embodiment, the front pulse neuron output information includes：Front pulse neuron is exported
Pulse tip formation, front pulse neuron and current PRF neuron connection weight index；
The current PRF neuronal messages include：Pulse tip formation sequence in current time window width, current time window
Row, history film potential information and film potential leakage information；
Then according to the front pulse neuron output information and the current PRF neuronal messages, current PRF is calculated
Neuron output information, including：
Connection weight according to the front pulse neuron and current PRF neuron is indexed, and reads front pulse nerve
Unit and the connection weight of current PRF neuron；
According to the pulse tip formation that the front pulse neuron is exported, and pulse tip letter in the current time window
Breath sequence, updates pulse tip formation sequence in the current time window, obtains pulse tip formation in current time window and updates
Sequence；
According to pulse tip formation renewal sequence in the current time window width, the current time window, by decay
Function calculates front pulse neuron input information；
The company of information, the front pulse neuron and current PRF neuron is input into according to the front pulse neuron
Weight, the history film potential information, film potential leakage information are connect, by spiking neuron computation model, calculates current
Spiking neuron output information.
It is described when the current PRF neuron output information is more than or equal to described adaptive wherein in one embodiment
When answering threshold value, the current PRF neuron output information is exported, and according to the first adaptive threshold updates model modification
Current adaptive threshold variable, also includes：
It is determined that providing trigger flag information to provide triggering, the granting trigger flag information includes providing triggering or granting
Do not trigger；
Reset refractory period timer, and the history film potential information is updated for default reset film potential information.
It is described wherein in one embodiment, the current PRF neuron output information is not exported, and according to second
Adaptive threshold updates current adaptive threshold variable described in model modification, also includes：
Determine that the granting trigger flag information is not triggered to provide；
Read the current time step of refractory period width and refractory period timer；
Whether the current time step according to the refractory period width and the refractory period timer, judge current time not
, by the refractory period timer one time step of cumulative timing, institute should not be updated in phase, if current time is in the refractory period
State history film potential information；
If current time is not in the refractory period, by the refractory period timer one time step of cumulative timing, and more
The new history film potential information is the current PRF neuron output information.
It is described to read current adaptive threshold variable and threshold potential wherein in one embodiment, and worked as according to described
Preceding adaptive threshold variable and the threshold potential, calculate current adaptive threshold, including：
Read random threshold value mask current potential, threshold bias, current adaptive threshold variable and random threshold value；
The random threshold value and the random threshold value mask current potential are carried out into step-by-step and operation, threshold value random superposition is obtained
Amount；
According to the threshold value random superposition amount and the threshold bias, the threshold potential is determined；
According to the threshold potential and the current adaptive threshold variable, the current adaptive threshold is determined.
Wherein in one embodiment, the adaptive thresholding current according to the first adaptive threshold updates model modification
Value variable, including：
Read and provide threshold delta and the current adaptive threshold variable；
According to default attenuation constant and the current adaptive threshold variable, first threshold is calculated；
The granting threshold delta is superimposed to the first threshold, Second Threshold is obtained；
The current adaptive threshold variable is updated according to the Second Threshold.
Wherein in one embodiment, the adaptive thresholding current according to the second adaptive threshold updates model modification
Value variable, including：
Read the current adaptive threshold variable；
According to the default attenuation constant and the current adaptive threshold variable, the 3rd threshold value is calculated,
The current adaptive threshold variable is updated according to the 3rd threshold value.
Wherein in one embodiment, the output current PRF neuron output information, including：
Read to provide and enable mark, the granting enables mark to be included allowing to provide data or do not allow granting data, when
The granting enable is designated when allowing to provide data,
The granting trigger flag information is read, when the granting trigger flag information is to provide triggering；
Export the current PRF neuron output information.
Adaptive threshold neuronal messages processing method provided by the present invention, in current PRF neuron output information
In calculating, by reading current adaptive threshold variable and threshold potential, current adaptive threshold is calculated；According to current adaptive
Answer threshold value, it is determined whether output current PRF neuron output information, and determine to update the model of current adaptive dependent variable.This hair
The neuronal messages processing method of bright provided adaptive threshold, can cause currently to have provided the neuron threshold of output information
Value is raised, and next time provides difficulty to be increased；And the current neuron threshold value reduction for not providing output information, provide difficulty drop next time
It is low, the granting frequency of each neuron in efficient balance whole network so that each neuron can be sent out in processing information
The effect of waving, greatly improves the information processing capability of impulsive neural networks.
Wherein in one embodiment, according to the front pulse neuron export pulse tip formation, and it is described work as
Pulse tip formation sequence in preceding time window, updates pulse tip formation sequence in the current time window, obtains current time
Pulse tip formation renewal sequence in window, according to the current time window width, the front pulse neuron and current PRF
The connection weight of neuron, calculates front pulse neuron and is input into information by attenuation function, can support with time depth
Space-time Pulse neural network model, compared to time depth be only one nerual network technique scheme, can greatly improve
The space time information code capacity of impulsive neural networks, enriches the application space of impulsive neural networks.
Wherein in one embodiment, by reading random threshold value mask current potential and threshold bias, and configuration deposit is received
The Configuration Values that device is given, determine the threshold potential so that neuron provides pulse tip formation has the random of certain probability
Property, no matter film potential more than fixed threshold either with or without biasing, due to there is depositing for threshold value random superposition amount that can just can be being born
It is likely to provide pulse in, the pericaryon, improves the computing capability and information processing energy of impulsive neural networks model
Power
Wherein in one embodiment, mark and granting trigger flag are enabled by setting to provide, determine current PRF god
Through first output information so that the controllability of the output of spiking neuron is higher, the neuron that enabler flags can be configured with is provided
Granting data are not allowed, and is only used as middle auxiliary and is calculated neuron, this needs the work(of multi-neuron cooperation completion for some
Can be very important.
The present invention also provides a kind of adaptive threshold neuronal messages processing system, including：
Front pulse neuron output information receiver module, for receiving front-end spiking neuron output information；
Current PRF neuronal messages read module, for reading current PRF neuronal messages；
Current PRF neuron output information computing module, for according to the front pulse neuron output information and institute
Current PRF neuronal messages are stated, current PRF neuron output information is calculated；
Current adaptive threshold computing module, for reading current adaptive threshold variable and threshold potential, and according to institute
Current adaptive threshold variable and the threshold potential are stated, current adaptive threshold is calculated；
Current PRF neuron output information output module, for whether judging the current PRF neuron output information
More than or equal to the adaptive threshold, if so, the current PRF neuron output information is then exported, and it is adaptive according to first
Threshold value is answered to update current adaptive threshold variable described in model modification, if it is not,
Do not export the current PRF neuron output information then, and model modification institute is updated according to the second adaptive threshold
State current adaptive threshold variable.
Do not export the current PRF neuron output information then, and model modification institute is updated according to the second adaptive threshold
Current adaptive threshold variable is stated wherein in one embodiment, the front pulse neuron output information includes：Front end arteries and veins
Rush the connection weight index of the pulse tip formation, front pulse neuron and current PRF neuron of neuron output；
The current PRF neuronal messages include：Pulse tip formation sequence in current time window width, current time window
Row, history film potential information and film potential leakage information；
The current PRF neuron output information computing module, including：
Spiking neuron connection weight reading unit, for according to the front pulse neuron and current PRF neuron
Connection weight index, read the connection weight of front pulse neuron and current PRF neuron；
Pulse tip formation sequence updating block in time window, for the pulse exported according to the front pulse neuron
Pulse tip formation sequence in tip formation, and the current time window, updates pulse tip formation in the current time window
Sequence, obtains pulse tip formation renewal sequence in current time window；
Front pulse neuron is input into information calculating unit, for according to the current time window width, the front end arteries and veins
The connection weight of neuron and current PRF neuron is rushed, calculating front pulse neuron by attenuation function is input into information；
Spiking neuron output information computing unit, for according to the front pulse neuron be input into information, it is described work as
Pulse tip formation renewal sequence, the history film potential information, film potential leakage information in preceding time window, by pulse
Neural relationship, calculates current PRF neuron output information.
Wherein in one embodiment, also include：
Triggering determining unit is provided, for determining that provide trigger flag information triggers to provide, the granting trigger flag
Information includes that providing triggering or granting does not trigger；
Trigger action unit is provided, for refractory period timer, and it is default multiple to update the history film potential information
Position film potential information.
Wherein in one embodiment, also include：
When provide triggering determining unit, it is determined that the granting trigger flag information for provide do not trigger；
Not trigger action unit is provided, the current time for reading refractory period width and refractory period timer is walked；According to
Whether the current time step of the refractory period width and the refractory period timer, judges current time in refractory period, if working as
The preceding time in the refractory period, by the refractory period timer one time step of cumulative timing, the history film electricity is not updated
Position information；If current time is not in the refractory period, by the refractory period timer one time step of cumulative timing, and update
The history film potential information is the current PRF neuron output information.
Wherein in one embodiment, the current adaptive threshold computing module, including：
Threshold information receiving unit, for reading random threshold value mask current potential, threshold bias and random threshold value；
Threshold value random superposition amount acquiring unit, for by the random threshold value and the random threshold value mask current potential carry out by
Position and operation, obtain threshold value random superposition amount；
Threshold potential determining unit, for according to the threshold value random superposition amount and the threshold bias, determining the threshold
Value current potential；
Current adaptive threshold determining unit, for according to the threshold potential and the current adaptive threshold variable,
Determine the current adaptive threshold.
Wherein in one embodiment, the current PRF neuron output information output module, including：
First adaptive threshold updating block, threshold delta and the current adaptive threshold variable are provided for reading；
According to default attenuation constant and the current adaptive threshold variable, first threshold is calculated；The granting threshold delta is folded
The first threshold is added to, Second Threshold is obtained；The current adaptive threshold variable is updated according to the Second Threshold.
Wherein in one embodiment, the current PRF neuron output information output module, including：
Second adaptive threshold updating block, for reading the current adaptive threshold variable；According to described default
Attenuation constant and the current adaptive threshold variable, calculate the 3rd threshold value, according to the 3rd threshold value update it is described it is current from
Adapt to thresholding variables.
Wherein in one embodiment, the current PRF neuron output information output module, including：
Provide and enable mark reading unit, provided for reading and enable mark, the enable mark of providing includes allowing hair
Put data or do not allow to provide data, when the granting enable is designated allows to provide data,
Trigger flag Information reading unit is provided, for reading the granting trigger flag information, triggering is provided when described
When flag information is triggered to provide；
Current PRF neuron output information output unit, for exporting the current PRF neuron output information.
Adaptive threshold neuronal messages processing system provided by the present invention, in current PRF light nerve output information
In calculating, by reading current adaptive threshold variable and threshold potential, current adaptive threshold is calculated；According to current adaptive
Answer threshold value, it is determined whether output current PRF neuron output information, and determine to update the model of current adaptive dependent variable.This hair
The neuronal messages processing system of bright provided adaptive threshold, can cause that the neuron threshold value of current granting is raised, under
Secondary granting difficulty increases；And the current neuron threshold value reduction do not provided, provide difficulty reduction next time.In this way, can be effectively equal
The granting frequency of each neuron in weighing apparatus whole network so that each neuron can play a role in processing information, greatly
The big information processing capability for improving impulsive neural networks.
Wherein in one embodiment, according to the front pulse neuron export pulse tip formation, and it is described work as
Pulse tip formation sequence in preceding time window, updates pulse tip formation sequence in the current time window, obtains current time
Pulse tip formation renewal sequence in window, according to the current time window width, the front pulse neuron and current PRF
The connection weight of neuron, calculates front pulse neuron and is input into information by attenuation function, can support with time depth
Space-time Pulse neural network model, compared to time depth be only one nerual network technique scheme, can greatly improve
The space time information code capacity of impulsive neural networks, enriches the application space of impulsive neural networks.
Wherein in one embodiment, by reading random threshold value mask current potential and threshold bias, and configuration deposit is received
The Configuration Values that device is given, determine the threshold potential so that neuron provides pulse tip formation has the random of certain probability
Property, no matter film potential more than fixed threshold either with or without biasing, due to there is depositing for threshold value random superposition amount that can just can be being born
It is likely to provide pulse in, the pericaryon, improves the computing capability and information processing energy of impulsive neural networks model
Power.
Wherein in one embodiment, mark and granting trigger flag are enabled by setting to provide, determine current PRF god
Through first output information so that the controllability of the output of spiking neuron is higher, the neuron that enabler flags can be configured with is provided
Granting data are not allowed, and is only used as middle auxiliary and is calculated neuron, this needs the work(of multi-neuron cooperation completion for some
Can be very important.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the adaptive threshold neutral net information processing method of one embodiment；
Fig. 2 is the schematic flow sheet of the adaptive threshold neutral net information processing method of another embodiment；
Fig. 3 is the schematic flow sheet of the adaptive threshold neutral net information processing method of another embodiment；
Fig. 4 is the schematic flow sheet of the adaptive threshold neutral net information processing method of further embodiment；
Fig. 5 is the schematic flow sheet of the adaptive threshold neutral net information processing method of further embodiment；
Fig. 6 is the schematic flow sheet of the adaptive threshold neutral net information processing method of further embodiment；
Fig. 7 is the structural representation of the adaptive threshold neutral net information processing system of one embodiment；
Fig. 8 is the structural representation of the adaptive threshold neutral net information processing system of another embodiment；
Fig. 9 is the structural representation of the adaptive threshold neutral net information processing system of another embodiment；
Figure 10 is the structural representation of the adaptive threshold neutral net information processing method, system of further embodiment.
Specific embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, below in conjunction with drawings and Examples pair
The present invention is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the present invention, not
For limiting the present invention.
Fig. 1 is the schematic flow sheet of the adaptive threshold neutral net information processing method of one embodiment, as shown in Figure 1
Adaptive threshold neutral net information processing method, including：
Step S100, receiving front-end spiking neuron output information.
Specifically, the front pulse neuron output information, is the arteries and veins of the front end being connected with current PRF neuron
Rush the pulse information of neuron output.
Step S200, reads current PRF neuronal messages.
Specifically, the current PRF neuronal messages, including the front pulse neuron of Current neural unit storage sends
History pulse information sequence etc..
Step S300, according to the front pulse neuron output information and the current PRF neuronal messages, calculates
Current PRF neuron output information.
Specifically, the pulse information that current PRF neuron is exported according to the front pulse neuron for receiving, and read
The current PRF neuronal messages for arriving, calculate current PRF neuron output information.
Step S400, reads current adaptive threshold variable and threshold potential, and become according to the current adaptive threshold
Amount and the threshold potential, calculate current adaptive threshold.
Specifically, the adaptive threshold variable, calculates according to a time step on current PRF neuron
Current PRF neuron output information (i.e. spike), if finally sent what is determined.If the pulse information of a upper time step
Do not sent, then the adaptive threshold variable continues decay on the basis of a upper time step, so that next time step
Pulse train granting difficulty reduction, if the pulse information of a upper time step is sent, the adaptive threshold variable
After decaying on the basis of a upper time step, increase a fixed increment, so that the pulse train of next time step
Providing difficulty increases.
Step S500, when whether the current PRF neuron output information is more than or equal to the adaptive threshold,
If so, then meeting step S600, step S700 is otherwise skipped to.
Specifically, whether current PRF neuron output information is judged more than or equal to the adaptive threshold, with after an action of the bowels
Continuous step determines whether to send the current PRF neuron output information according to judged result.
Step S600, exports the current PRF neuron output information, and according to the first adaptive threshold more new model
Update the current adaptive threshold variable.
If specifically, current PRF neuron output information is more than or equal to the adaptive threshold, sending described working as
Prepulse neuron output information, current adaptive threshold becomes according to first adaptive threshold updates model modification
Amount, so that the current adaptive threshold variable is in the calculating of next time step, increases a fixed increment, under increase
The granting difficulty of the current PRF neuron output information that one time step is calculated.
Step S700, then do not export the current PRF neuron output information, and update according to the second adaptive threshold
Current adaptive threshold variable described in model modification.
If specifically, current PRF neuron output information is less than the adaptive threshold, the current arteries and veins is not sent
Neuron output information is rushed, the current adaptive threshold variable according to second adaptive threshold updates model modification, with
Make the current adaptive threshold variable in the calculating of next time step, continue to decay, reduce next time step and calculate
The granting difficulty of the current PRF neuron output information for going out.
Adaptive threshold neuronal messages processing method provided by the present invention, in current PRF neuron output information
In calculating, by reading current adaptive threshold variable and threshold potential, current adaptive threshold is calculated；According to current adaptive
Answer threshold value, it is determined whether output current PRF neuron output information, and determine to update the model of current adaptive dependent variable.This hair
The spiking neuron information processing method of bright provided adaptive threshold, can cause currently to have provided the nerve of output information
First threshold value is raised, and next time provides difficulty to be increased；And the current neuron threshold value reduction for not providing output information, provide difficulty next time
Reduce, the granting frequency of each neuron in efficient balance whole network so that each neuron can be in processing information
Play a role, greatly improve the information processing capability of impulsive neural networks.
Fig. 2 is the schematic flow sheet of the adaptive threshold neutral net information processing method of another embodiment, such as Fig. 2 institutes
The adaptive threshold neutral net information processing method for showing, including：
Step S100b, pulse tip formation, front pulse neuron and current arteries and veins that receiving front-end spiking neuron is exported
Rush the connection weight index of neuron.
Specifically, the front pulse neuron is indexed with the connection weight of current PRF neuron, it is front end neuron
The weight index together sent with the front pulse neuron output information, the extraction for indicating Current neural unit weight.
The pulse tip formation of the front pulse neuron output, is the pulse tip signal of front pulse neuron transmission
(spike)。
Step S200b, reads pulse tip formation sequence, history film potential in current time window width, current time window
Information and film potential leakage information.
Specifically, pulse tip formation sequence in the current time window, refers in the current time window width, to incite somebody to action
The pulse tip formation that past a range of time step is received, the information sequence for caching successively in chronological order.
Step S300b, the connection weight according to the front pulse neuron and current PRF neuron is indexed, before reading
The connection weight of end spiking neuron and current PRF neuron.
Specifically, the front pulse neuron is indexed with the connection weight of current PRF neuron, it is an address letter
Breath, Current neural unit indexes according to the front pulse neuron for receiving with the connection weight of current PRF neuron,
In memory in Current neural unit, the connection weight of front pulse neuron and current PRF neuron is read, according to institute
The connection weight information stated, can participate in the calculating of Current neural unit output information by the output information of front end neuron
Cheng Zhong, more accurately reflects the weight of the output information of front end neuron, carries more rich information.
Step S400b, according to the pulse tip formation that the front pulse neuron is exported, and in the current time window
Pulse tip formation sequence, updates pulse tip formation sequence in the current time window, obtains pulse point in current time window
Client information renewal sequence.
Specifically, the pulse tip formation sequence, walks in the operation of each spiking neuron, one is stored in sequence head
After new pulse tip formation, the pulse tip formation on a sequence tail position is deleted, update once pulse tip sequence
Row.
Step S500b, sequence is updated according to pulse tip formation in the current time window width, the current time window
Row, calculate front pulse neuron and are input into information by attenuation function.
Specifically, utilizingThe front pulse neuron input information is calculated, wherein, Tw
It is the time window width, δjAfter providing spike in current time window for front end neuron j, in the current time window
Time step in pulse tip formation renewal sequence.(Δ t) is an attenuation function to K, is reduced rapidly as Δ t increases.
Step S600b, information, the front pulse neuron and current PRF are input into according to the front pulse neuron
The connection weight of neuron, the history film potential information, film potential leakage information, mould is calculated by spiking neuron
Type, calculates current PRF neuron output information.
Specifically, representing that front pulse neuron is input into the calculating of information using equation below：
Wherein WijIt is the front pulse neuron j and the connection weight of current PRF neuron i, TwIt is the time window
Width, δjAfter spike being provided for front end neuron j in current time window, the pulse tip formation in the current time window
Time step in renewal sequence.(Δ t) is an attenuation function to K, is reduced rapidly as Δ t increases.VleakFor leakage value is believed
Breath.Basic model at cell space can be reduced to：
VSNN=f (V+Vinput+Vleak)
Provide model and reset model be constant, wherein V is the history film potential information that memory is preserved,
VinputIt is the current input clapped and add up, is equivalent to above-mentioned
VleakIt is leakage value information.
In the present embodiment, the pulse tip formation for being exported according to the front pulse neuron, and the current time
Pulse tip formation sequence in window, updates pulse tip formation sequence in the current time window, obtains arteries and veins in current time window
Tip formation renewal sequence is rushed, according to the current time window width, the front pulse neuron and current PRF neuron
Connection weight, front pulse neuron is calculated by attenuation function and is input into information, the space-time with time depth can be supported
Impulsive neural networks model, compared to the nerual network technique scheme that time depth is only, can greatly improve pulse god
Through the space time information code capacity of network, the application space of impulsive neural networks is enriched.
Fig. 3 is the schematic flow sheet of the adaptive threshold neutral net information processing method of another embodiment, such as Fig. 3 institutes
The adaptive threshold neutral net information processing method for showing, including：
Step S100c, calculates current PRF neuron output information and adaptive threshold.
Whether step S200c, judge the current PRF neuron output information more than or equal to the adaptive threshold
When, determining that provide trigger flag information triggers to provide according to the comparative result, the granting trigger flag information includes hair
Put triggering or granting is not triggered, when it is determined that providing trigger flag information to provide triggering, meet step S300c, when it is determined that providing
When trigger flag information is not triggered to provide, step S400c is skipped to.
Specifically, according to the adaptive threshold current potential, be compared with the current PRF neuron output information, and
Determined to provide trigger flag information according to comparative result.Only described current PRF neuron output information is more than the self adaptation
During threshold potential, the current PRF neuron output information can just be sent.
Step S300c, reset refractory period timer, and the history film potential information is updated for default reset film potential
Information.
Specifically, when the granting trigger flag information is to provide triggering, the current PRF neuron output information
Sent, after refractory period timer is reset, recalculated refractory period, and update the history film potential information for default film
Electrical potential information, and described history film potential information updating, according to the reset types of configuration, be reset to for film potential to work as by selectivity
Preceding film potential, current film potential and threshold potential difference, or fixed reset voltage.
Step S400c, reads the current time step of refractory period width and refractory period timer.
Specifically, when the granting trigger flag information is not triggered to provide, the current PRF neuron output letter
Whether breath is not sent, determine whether current in refractory period.The refractory period width is the duration scope of refractory period, described
Refractory period timer utilizes the mode timing of time step.
Step S500c, the current time step according to the refractory period width and the refractory period timer, when judging current
Between whether in refractory period, if current time is in the refractory period, meet step S600c, otherwise skip to step S700c.
Specifically, the cumulative calculation of the current time step according to the refractory period timer, it can be determined that go out current time
Whether step is also in refractory period.
Step S600c, by the refractory period timer one time step of cumulative timing, the history film potential letter is not updated
Breath.
If specifically, current time is in the refractory period, according to the bionical feature of impulsive neural networks, not to the arteries and veins
The neural output information of punching carries out any response, not more new historical film potential information, the history film potential information, when being next
The spiking neuron of spacer step needs the information for reading, i.e., in refractory period, the spiking neuron output information that this is calculated is not
Participate in the calculating of next time step.
Step S700c, by the refractory period timer one time step of cumulative timing, and updates the history film potential letter
It is the current PRF neuron output information to cease.
Then it is the current PRF neuron output information by the history film potential information specifically, such as not answering outer,
Participate in the calculating of next time step.
In the present embodiment, by adaptive threshold current potential so that when neuron provides pulse tip formation with upper one
Whether spacer step has provided the current PRF neuron output information correlation, can be with each neuron in efficient balance whole network
Granting frequency so that each neuron can play a role in processing information, greatly improve the letter of impulsive neural networks
Breath disposal ability.
Wherein in one embodiment, the output current PRF neuron output information, including reading granting makes
Can identify, the enable mark of providing includes allowing granting data or do not allow to provide data, when the enable of providing is designated
When allowing to provide data, the granting trigger flag information is read, when the granting trigger flag information is to provide triggering；It is defeated
Go out the current PRF neuron output information.
In the present embodiment, mark and granting trigger flag are enabled by setting granting, determines that current PRF neuron is defeated
Go out information so that the controllability of the output of spiking neuron is higher, the neuron that providing enabler flags can be configured with is not allowed
Granting data, and be only used as middle auxiliary and calculate neuron, this needs the function right and wrong of multi-neuron cooperation completion for some
It is often necessary.
Fig. 4 is the schematic flow sheet of the adaptive threshold neutral net information processing method of further embodiment, such as Fig. 4 institutes
The adaptive threshold neutral net information processing method for showing, including：
Step S410, reads random threshold value mask current potential, threshold bias, current adaptive threshold variable and random threshold value.
Specifically, the current adaptive threshold variable, is when a task starts, according to a default self adaptation
Threshold value initial value, default attenuation coefficient and provide threshold delta, progressively decay or raise according to each time step, draw from
Adapt to thresholding variables value.The current adaptive threshold variable, be current PRF neuron read current time step from
Adapt to thresholding variables.The random threshold value provides a random number seed by configuration register, by pseudorandom number generator
Produce.
Step S420, step-by-step and operation are carried out by the random threshold value and the random threshold value mask current potential, obtain threshold value
Random superposition amount.
Specifically, the random threshold value mask current potential is used for the scope of threshold limit increment.
Step S430, according to the threshold value random superposition amount and the threshold bias, determines the threshold potential.
Specifically, the threshold value random superposition amount is added with the threshold bias, the threshold potential is obtained.
Step S440, according to the threshold potential and the current adaptive threshold variable, determines the current self adaptation
Threshold value.
Specifically, by the threshold potential and the current adaptive threshold addition of variables, determining the current self adaptation
Threshold value.Wherein, by the threshold value random superposition amount and default threshold bias Vth0It is added, produces real threshold potential Vth.Its
In, the seed of pseudorandom number generator is by configuration register VseedBe given.Mask current potential VmaskFor the model of threshold limit increment
Enclose：If Vmask=0, then threshold value random superposition amount is also 0, and Firing Patterns deteriorate to fixed threshold granting, and fixed threshold is Vth0；
If Vmask≠ 0, then Firing Patterns are the granting of part probability threshold value.As extreme case Vth0=0, then Firing Patterns are full-probability
Threshold value is provided.
In the present embodiment, be given by reading random threshold value mask current potential and threshold bias, and receive configuration register
Configuration Values, determine the threshold potential so that neuron provides pulse tip formation has the randomness of certain probability no matter
Film potential is biased either with or without more than fixed threshold, due to also having a presence for the threshold value random superposition amount that can just can be being born, the god
It is likely to provide pulse through first cell space, improves the computing capability and information processing capability of impulsive neural networks model.
Fig. 5 is the schematic flow sheet of the adaptive threshold neutral net information processing method of further embodiment, such as Fig. 5 institutes
The adaptive threshold neutral net information processing method for showing, in being Fig. 1, according to the first adaptive threshold updates model modification
The detailed step of current adaptive threshold variable part, including：
Step S610, reads and provides threshold delta and the current adaptive threshold variable.
Specifically, the granting threshold delta is default value, can be set according to the demand of task.
Step S620, according to default attenuation constant and the current adaptive threshold variable, calculates first threshold.
Specifically, the current adaptive threshold variable that will be read, calculates according to default attenuation coefficient, the first threshold is obtained
Value, the first threshold is the adaptive threshold variable after this time step performs decay.
Step S630, the first threshold is superimposed to by the granting threshold delta, obtains Second Threshold.
Specifically, after this time step performs decay, then by first described in the default granting threshold delta superposition value
Threshold value, obtains Second Threshold.
Step S640, the current adaptive threshold variable is updated according to the Second Threshold.
Specifically, being the Second Threshold by the current adaptive threshold variable update.Because of the Second Threshold, be
After adaptive dependent variable performs decay calculation, default granting threshold delta has been superimposed again so that what next time step read
Adaptive threshold increases, and improves the granting difficulty of the current PRF neuron output information of next time step.
The first adaptive threshold more new model provided in the present embodiment, sends in current PRF neuron output information
In the case of, by current adaptive threshold perform decay after, be superimposed a default increment so that next time step it is current
The granting difficulty of spiking neuron output information increases.In this way for threshold value adds flexible characteristic, can cause whole
Than more uniform, each neuron has an opportunity to learn respective input receptive field the granting rate of network.
Fig. 6 is the schematic flow sheet of the adaptive threshold neutral net information processing method of further embodiment, such as Fig. 6 institutes
The adaptive threshold neutral net information processing method for showing, in being Fig. 1, according to the second adaptive threshold updates model modification
The detailed step of current adaptive threshold variable part, including：
Step S710, reads the current adaptive threshold variable.
Specifically, because current PRF neuron output information is not sent, then adaptive threshold need not be improved.
Step S720, according to the default attenuation constant and the current adaptive threshold variable, calculates the 3rd threshold value.
Specifically, the current adaptive threshold variable that will be read, calculates according to default attenuation coefficient, the 3rd threshold is obtained
Value, the 3rd threshold value is the adaptive threshold variable after this time step performs decay.
Step S730, the current adaptive threshold variable is updated according to the 3rd threshold value.
Specifically, the adaptive dependent variable after by decay, i.e. the 3rd threshold value are set as current adaptive threshold increment, reduce
The granting difficulty of the current PRF neuron output information of next time step.
The second adaptive threshold more new model provided in the present embodiment, does not send out in current PRF neuron output information
In the case of sending, current adaptive threshold is performed into decay so that the current PRF neuron output information of next time step
Granting difficulty reduction.In this way for threshold value adds flexible characteristic, can cause that the granting rate of whole network is more equal
Even, each neuron has an opportunity to learn respective input receptive field.
Fig. 7 is the structural representation of the adaptive threshold neutral net information processing system of one embodiment, as shown in Figure 7
Adaptive threshold neutral net information processing system, including：
Front pulse neuron output information receiver module 100, for receiving front-end spiking neuron output information；It is described
Front pulse neuron output information includes：Front pulse neuron output pulse tip formation, front pulse neuron with
The connection weight index of current PRF neuron.
Current PRF neuronal messages read module 200, for reading current PRF neuronal messages；The current PRF
Neuronal messages include：Pulse tip formation sequence, history film potential information and film in current time window width, current time window
Potential leakage information.
Current PRF neuron output information computing module 300, for according to the front pulse neuron output information
With the current PRF neuronal messages, current PRF neuron output information is calculated；
Current adaptive threshold computing module 400, for reading current adaptive threshold variable and threshold potential, and according to
The current adaptive threshold variable and the threshold potential, calculate current adaptive threshold；
Current PRF neuron output information output module 500, for when the current PRF neuron output information it is big
When the adaptive threshold, the current PRF neuron output information is exported, and according to the first adaptive threshold
Current adaptive threshold variable described in model modification is updated, the current PRF neuron output information is not exported then otherwise, and
The current adaptive threshold variable according to the second adaptive threshold updates model modification.
Adaptive threshold neuronal messages processing system provided by the present invention, in current PRF light nerve output information
In calculating, by reading current adaptive threshold variable and threshold potential, current adaptive threshold is calculated；According to current adaptive
Answer threshold value, it is determined whether output current PRF neuron output information, and determine to update the model of current adaptive dependent variable.This hair
The neuronal messages processing system of bright provided adaptive threshold, can cause that the neuron threshold value of current granting is raised, under
Secondary granting difficulty increases；And the current neuron threshold value reduction do not provided, provide difficulty reduction next time.In this way, can be effectively equal
The granting frequency of each neuron in weighing apparatus whole network so that each neuron can play a role in processing information, greatly
The big information processing capability for improving impulsive neural networks.
Fig. 8 is the structural representation of the adaptive threshold neutral net information processing system of another embodiment, such as Fig. 8 institutes
The adaptive threshold neutral net information processing system shown, is the output information computing module of current PRF neuron described in Fig. 7
300, including：
Spiking neuron connection weight reading unit 100b, for according to the front pulse neuron and current PRF god
Indexed through the connection weight of unit, read the connection weight of front pulse neuron and current PRF neuron；
Pulse tip formation sequence updating block 200b in time window, for what is exported according to the front pulse neuron
Pulse tip formation sequence in pulse tip formation, and the current time window, updates pulse tip in the current time window
Information sequence, obtains pulse tip formation renewal sequence in current time window；
Front pulse neuron be input into information calculating unit 300b, for according to the current time window width, it is described before
End spiking neuron and the connection weight of current PRF neuron, calculate front pulse neuron and are input into letter by attenuation function
Breath；
Spiking neuron output information computing unit 400b, for being input into information, institute according to the front pulse neuron
Pulse tip formation renewal sequence in current time window, the history film potential information, film potential leakage information are stated, is passed through
Spiking neuron computation model, calculates current PRF neuron output information.
In the present embodiment, the pulse tip formation for being exported according to the front pulse neuron, and the current time
Pulse tip formation sequence in window, updates pulse tip formation sequence in the current time window, obtains arteries and veins in current time window
Tip formation renewal sequence is rushed, according to the current time window width, the front pulse neuron and current PRF neuron
Connection weight, front pulse neuron is calculated by attenuation function and is input into information, the space-time with time depth can be supported
Impulsive neural networks model, compared to the nerual network technique scheme that time depth is only, can greatly improve pulse god
Through the space time information code capacity of network, the application space of impulsive neural networks is enriched.
Fig. 9 is the structural representation of the adaptive threshold neutral net information processing system of another embodiment, such as Fig. 9 institutes
The adaptive threshold neutral net information processing system shown, is the output information output module of current PRF neuron described in Fig. 7,
Including：
Triggering determining unit 100c is provided, it is described to provide triggering for determining that provide trigger flag information triggers to provide
Flag information includes that providing triggering or granting does not trigger；
Trigger action unit 200c is provided, for refractory period timer, and it is default to update the history film potential information
Reset film potential information.
When provide triggering determining unit 100c, it is determined that the granting trigger flag information for provide do not trigger；
Not trigger action unit 300c is provided, the current time for reading refractory period width and refractory period timer is walked；
Whether current time step according to the refractory period width and the refractory period timer, judges current time in refractory period,
If current time is in the refractory period, by the refractory period timer one time step of cumulative timing, the history is not updated
Film potential information；If current time is not in the refractory period, by the refractory period timer one time step of cumulative timing, and
The history film potential information is updated for the current PRF neuron output information.
First adaptive threshold updating block 400c, threshold delta and the current adaptive threshold change are provided for reading
Amount；According to default attenuation constant and the current adaptive threshold variable, first threshold is calculated；By the granting threshold delta
The first threshold is superimposed to, Second Threshold is obtained；The current adaptive threshold variable is updated according to the Second Threshold.
Second adaptive threshold updating block 500c, for reading the current adaptive threshold variable；According to described pre-
If attenuation constant and the current adaptive threshold variable, calculate the 3rd threshold value, updated according to the 3rd threshold value it is described ought
Preceding adaptive threshold variable.
Provide and enable mark reading unit 600c, provided for reading and enable mark, the enable mark of providing includes permitting
Perhaps data are provided or does not allow to provide data, when the granting enable is designated allows to provide data,
Trigger flag Information reading unit 700c is provided, for reading the granting trigger flag information, when the granting
When trigger flag information is triggered to provide.
Current PRF neuron output information output unit 800c, for exporting the current PRF neuron output letter
Breath.
Adaptive threshold more new model provided in the present embodiment, is sent in current PRF neuron output information
In the case of, after current adaptive threshold is performed into decay, it is superimposed a default increment so that the current arteries and veins of next time step
The granting difficulty for rushing neuron output information increases.In the case where current PRF neuron output information does not send, will be current
Adaptive threshold performs decay so that the granting difficulty reduction of the current PRF neuron output information of next time step.It is logical
This mode is crossed for threshold value adds flexible characteristic, the granting rate of whole network can be caused than more uniform, each neuron has
Chance learns respective input receptive field.
Figure 10 is the structural representation of the adaptive threshold neutral net information processing method, system of further embodiment, is such as schemed
Adaptive threshold neutral net information processing system shown in 10, is the current adaptive threshold computing module 400 in Fig. 7, bag
Include：
Threshold information receiving unit 410, for reading random threshold value mask current potential, threshold bias and random threshold value；
Threshold value random superposition amount acquiring unit 420, for the random threshold value and the random threshold value mask current potential to be entered
Row step-by-step and operation, obtain threshold value random superposition amount；
Threshold potential determining unit 430, for according to the threshold value random superposition amount and the threshold bias, it is determined that described
Threshold potential；
Current adaptive threshold determining unit 440, for being become according to the threshold potential and the current adaptive threshold
Amount, determines the current adaptive threshold.
In the present embodiment, be given by reading random threshold value mask current potential and threshold bias, and receive configuration register
Configuration Values, determine the threshold potential so that neuron provides pulse tip formation has the randomness of certain probability no matter
Film potential is biased either with or without more than fixed threshold, due to also having a presence for the threshold value random superposition amount that can just can be being born, the god
It is likely to provide pulse through first cell space, improves the computing capability and information processing capability of impulsive neural networks model.
Each technical characteristic of embodiment described above can be combined arbitrarily, to make description succinct, not to above-mentioned reality
Apply all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, the scope of this specification record is all considered to be.
Embodiment described above only expresses several embodiments of the invention, and its description is more specific and detailed, but simultaneously
Can not therefore be construed as limiting the scope of the patent.It should be pointed out that coming for one of ordinary skill in the art
Say, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to protection of the invention
Scope.Therefore, the protection domain of patent of the present invention should be determined by the appended claims.
Claims (16)
1. a kind of adaptive threshold neuronal messages processing method, it is characterised in that methods described includes：
Receiving front-end spiking neuron output information；
Read current PRF neuronal messages；
According to the front pulse neuron output information and the current PRF neuronal messages, current PRF neuron is calculated
Output information；
Current adaptive threshold variable and threshold potential are read, and according to the current adaptive threshold variable and threshold value electricity
Position, calculates current adaptive threshold；
Whether the current PRF neuron output information is judged more than or equal to the adaptive threshold, if so, then exporting institute
Current PRF neuron output information is stated, and current adaptive threshold becomes according to the first adaptive threshold updates model modification
Amount, if it is not,
The current PRF neuron output information is not exported, and it is current according to the second adaptive threshold updates model modification
Adaptive threshold variable.
2. adaptive threshold neuronal messages processing method according to claim 1, it is characterised in that：
The front pulse neuron output information includes：Pulse tip formation, the front pulse of front pulse neuron output
Neuron is indexed with the connection weight of current PRF neuron；
The current PRF neuronal messages include：Pulse tip formation sequence in current time window width, current time window, go through
History film potential information and film potential leakage information；
Then according to the front pulse neuron output information and the current PRF neuronal messages, current PRF nerve is calculated
First output information, including：
Connection weight according to the front pulse neuron and current PRF neuron is indexed, read front pulse neuron with
The connection weight of current PRF neuron；
According to the pulse tip formation that the front pulse neuron is exported, and pulse tip formation sequence in the current time window
Row, update pulse tip formation sequence in the current time window, obtain pulse tip formation renewal sequence in current time window；
According to pulse tip formation renewal sequence in the current time window width, the current time window, by attenuation function
Calculate front pulse neuron input information；
The connection weight of information, the front pulse neuron and current PRF neuron is input into according to the front pulse neuron
Weight, the history film potential information, the film potential leakage information, by spiking neuron computation model, calculate current PRF
Neuron output information.
3. adaptive threshold neuronal messages processing method according to claim 2, it is characterised in that described to work as when described
When prepulse neuron output information is more than or equal to the adaptive threshold, the current PRF neuron output letter is exported
Breath, and the current adaptive threshold variable according to the first adaptive threshold updates model modification, also include：
It is determined that providing trigger flag information to provide triggering, the granting trigger flag information includes providing triggering or providing not touching
Hair；
Reset refractory period timer, and the history film potential information is updated for default reset film potential information.
4. adaptive threshold neuronal messages processing method according to claim 3, it is characterised in that described, does not export
The current PRF neuron output information, and the current adaptive threshold according to the second adaptive threshold updates model modification
Variable, also includes：
Determine that the granting trigger flag information is not triggered to provide；
Read the current time step of refractory period width and refractory period timer；
Whether the current time step according to the refractory period width and the refractory period timer, judge current time in refractory period
It is interior, if current time is in the refractory period, by the refractory period timer one time step of cumulative timing, described going through is not updated
History film potential information；
If current time is not in the refractory period, by the refractory period timer one time step of cumulative timing, and institute is updated
History film potential information is stated for the current PRF neuron output information.
5. adaptive threshold neuronal messages processing method according to claim 1, it is characterised in that the reading is current
Adaptive threshold variable and threshold potential, and according to the current adaptive threshold variable and the threshold potential, calculate current
Adaptive threshold, including：
Read random threshold value mask current potential, threshold bias, current adaptive threshold variable and random threshold value；
The random threshold value and the random threshold value mask current potential are carried out into step-by-step and operation, threshold value random superposition amount is obtained；
According to the threshold value random superposition amount and the threshold bias, the threshold potential is determined；
According to the threshold potential and the current adaptive threshold variable, the current adaptive threshold is determined.
6. adaptive threshold neuronal messages processing method according to claim 1, it is characterised in that described according to first
Adaptive threshold updates current adaptive threshold variable described in model modification, including：
Read and provide threshold delta and the current adaptive threshold variable；
According to default attenuation constant and the current adaptive threshold variable, first threshold is calculated；
The granting threshold delta is superimposed to the first threshold, Second Threshold is obtained；
The current adaptive threshold variable is updated according to the Second Threshold.
7. adaptive threshold neuronal messages processing method according to claim 6, it is characterised in that described according to second
Adaptive threshold updates current adaptive threshold variable described in model modification, including：
Read the current adaptive threshold variable；
According to the default attenuation constant and the current adaptive threshold variable, the 3rd threshold value is calculated,
The current adaptive threshold variable is updated according to the 3rd threshold value.
8. adaptive threshold neuronal messages processing method according to claim 3, it is characterised in that described in the output
Current PRF neuron output information, including：
Read to provide and enable mark, the granting enables mark to be included allowing to provide data or do not allow granting data, when described
Granting enable is designated when allowing to provide data,
The granting trigger flag information is read, when the granting trigger flag information is to provide triggering；
Export the current PRF neuron output information.
9. a kind of adaptive threshold neuronal messages processing system, it is characterised in that including：
Front pulse neuron output information receiver module, for receiving front-end spiking neuron output information；
Current PRF neuronal messages read module, for reading current PRF neuronal messages；
Current PRF neuron output information computing module, for according to the front pulse neuron output information and it is described work as
Prepulse neuronal messages, calculate current PRF neuron output information；
Current adaptive threshold computing module, for reading current adaptive threshold variable and threshold potential, and works as according to described
Preceding adaptive threshold variable and the threshold potential, calculate current adaptive threshold；
Current PRF neuron output information output module, for judging whether the current PRF neuron output information is more than
Or equal to the adaptive threshold, if so, the current PRF neuron output information is then exported, and according to the first adaptive thresholding
Value updates current adaptive threshold variable described in model modification, if it is not,
Do not export the current PRF neuron output information then, and according to the second adaptive threshold updates model modification when
Preceding adaptive threshold variable.
10. adaptive threshold neuronal messages processing system according to claim 9, it is characterised in that：
The front pulse neuron output information includes：Pulse tip formation, the front pulse of front pulse neuron output
Neuron is indexed with the connection weight of current PRF neuron；
The current PRF neuronal messages include：Pulse tip formation sequence in current time window width, current time window, go through
History film potential information and film potential leakage information；
The current PRF neuron output information computing module, including：
Spiking neuron connection weight reading unit, for the company according to the front pulse neuron and current PRF neuron
Weight index is connect, the connection weight of front pulse neuron and current PRF neuron is read；
Pulse tip formation sequence updating block in time window, for the pulse tip exported according to the front pulse neuron
Pulse tip formation sequence in information, and the current time window, updates pulse tip formation sequence in the current time window,
Obtain pulse tip formation renewal sequence in current time window；
Front pulse neuron is input into information calculating unit, for according to the current time window width, front pulse god
Through unit and the connection weight of current PRF neuron, front pulse neuron is calculated by attenuation function and is input into information；
Spiking neuron output information computing unit, for according to the front pulse neuron be input into information, it is described current when
Between pulse tip formation renewal sequence, the history film potential information, film potential leakage information in window, by pulse nerve
Relationship, calculates current PRF neuron output information.
11. adaptive threshold neuronal messages processing systems according to claim 10, it is characterised in that also include：
Triggering determining unit is provided, for determining that provide trigger flag information triggers to provide, the granting trigger flag information
Do not triggered including providing triggering or granting；
Trigger action unit is provided, for refractory period timer, and the history film potential information is updated for default reset film
Electrical potential information.
12. adaptive threshold neuronal messages processing systems according to claim 11, it is characterised in that also include：
When provide triggering determining unit, it is determined that the granting trigger flag information for provide do not trigger；
Not trigger action unit is provided, the current time for reading refractory period width and refractory period timer is walked；According to described
Whether the current time step of refractory period width and the refractory period timer, judges current time in refractory period, if when current
Between in the refractory period, by cumulative one time step of timing of the refractory period timer, the history film potential letter is not updated
Breath；If current time is not in the refractory period, by the refractory period timer one time step of cumulative timing, and update described
History film potential information is the current PRF neuron output information.
13. adaptive threshold neuronal messages processing systems according to claim 9, it is characterised in that it is described it is current from
Threshold calculation module is adapted to, including：
Threshold information receiving unit, for reading random threshold value mask current potential, threshold bias and random threshold value；
Threshold value random superposition amount acquiring unit, for by the random threshold value and the random threshold value mask current potential carry out step-by-step with
Operation, obtains threshold value random superposition amount；
Threshold potential determining unit, for according to the threshold value random superposition amount and the threshold bias, determining the threshold value electricity
Position；
Current adaptive threshold determining unit, for according to the threshold potential and the current adaptive threshold variable, it is determined that
The current adaptive threshold.
14. adaptive threshold neuronal messages processing systems according to claim 9, it is characterised in that the current arteries and veins
Neuron output information output module is rushed, including：
First adaptive threshold updating block, threshold delta and the current adaptive threshold variable are provided for reading；According to
Default attenuation constant and the current adaptive threshold variable, calculate first threshold；The granting threshold delta is superimposed to
The first threshold, obtains Second Threshold；The current adaptive threshold variable is updated according to the Second Threshold.
15. adaptive threshold neuronal messages processing systems according to claim 14, it is characterised in that described to work as
Prepulse neuron output information output module, including：
Second adaptive threshold updating block, for reading the current adaptive threshold variable；According to the default decay
Constant and the current adaptive threshold variable, calculate the 3rd threshold value, and the current self adaptation is updated according to the 3rd threshold value
Thresholding variables.
16. adaptive threshold neuronal messages processing systems according to claim 11, it is characterised in that the current arteries and veins
Neuron output information output module is rushed, including：
Provide and enable mark reading unit, provided for reading and enable mark, the enable mark of providing includes allowing to provide number
According to or do not allow provide data, when it is described granting enable be designated allow provide data when,
Trigger flag Information reading unit is provided, for reading the granting trigger flag information, when the granting trigger flag
When information is triggered to provide；
Current PRF neuron output information output unit, for exporting the current PRF neuron output information.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710042067.5A CN106875005B (en) | 2017-01-20 | 2017-01-20 | Adaptive threshold neuronal messages processing method and system |
| PCT/CN2017/114669 WO2018133570A1 (en) | 2017-01-20 | 2017-12-05 | Self-adaptive threshold neuron information processing method, self-adaptive leakage value neuron information processing method and system, and computer device and readable storage medium |
| US16/478,458 US11551074B2 (en) | 2017-01-20 | 2017-12-05 | Self-adaptive threshold neuron information processing method, self-adaptive leakage value neuron information processing method, system computer device and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710042067.5A CN106875005B (en) | 2017-01-20 | 2017-01-20 | Adaptive threshold neuronal messages processing method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106875005A true CN106875005A (en) | 2017-06-20 |
| CN106875005B CN106875005B (en) | 2019-09-20 |
Family
ID=59158522
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710042067.5A Active CN106875005B (en) | 2017-01-20 | 2017-01-20 | Adaptive threshold neuronal messages processing method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106875005B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107609640A (en) * | 2017-10-01 | 2018-01-19 | 胡明建 | A kind of threshold values selects the design method of end graded potential formula artificial neuron |
| WO2018133570A1 (en) * | 2017-01-20 | 2018-07-26 | 清华大学 | Self-adaptive threshold neuron information processing method, self-adaptive leakage value neuron information processing method and system, and computer device and readable storage medium |
| CN112734012A (en) * | 2021-01-07 | 2021-04-30 | 北京灵汐科技有限公司 | Impulse neural network training method, data processing method, electronic device, and medium |
| CN112766469A (en) * | 2021-01-22 | 2021-05-07 | 北京灵汐科技有限公司 | Method and device for generating pulse signal |
| CN112949833A (en) * | 2021-03-26 | 2021-06-11 | 北京航空航天大学 | Probability calculation neuron calculation unit and construction method |
| CN113449864A (en) * | 2021-07-21 | 2021-09-28 | 北京大学 | Feedback type pulse neural network model training method for image data classification |
| WO2022063202A1 (en) * | 2020-09-25 | 2022-03-31 | 北京灵汐科技有限公司 | Text classification method, apparatus, device, and storage medium |
| CN114861838A (en) * | 2022-07-06 | 2022-08-05 | 武汉理工大学 | Intelligent classification method for pulsatile neural brains based on neuron complex dynamics |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1078053A (en) * | 1992-04-27 | 1993-11-03 | 福建师范大学 | Injection electroluminescent nerve network circuit |
| CN102306301A (en) * | 2011-08-26 | 2012-01-04 | 中南民族大学 | Motion identification system by simulating spiking neuron of primary visual cortex |
| CN103430186A (en) * | 2012-01-20 | 2013-12-04 | 松下电器产业株式会社 | Learning method for neural network circuit |
| CN103778468A (en) * | 2014-01-16 | 2014-05-07 | 北京大学 | RRAM-based new type neural network circuit |
| CN105095967A (en) * | 2015-07-16 | 2015-11-25 | 清华大学 | Multi-mode neural morphological network core |
-
2017
- 2017-01-20 CN CN201710042067.5A patent/CN106875005B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1078053A (en) * | 1992-04-27 | 1993-11-03 | 福建师范大学 | Injection electroluminescent nerve network circuit |
| CN102306301A (en) * | 2011-08-26 | 2012-01-04 | 中南民族大学 | Motion identification system by simulating spiking neuron of primary visual cortex |
| CN103430186A (en) * | 2012-01-20 | 2013-12-04 | 松下电器产业株式会社 | Learning method for neural network circuit |
| CN103778468A (en) * | 2014-01-16 | 2014-05-07 | 北京大学 | RRAM-based new type neural network circuit |
| CN105095967A (en) * | 2015-07-16 | 2015-11-25 | 清华大学 | Multi-mode neural morphological network core |
Non-Patent Citations (3)
| Title |
|---|
| AAMIR ET.AL: "A highly tunable 65-nm CMOS LIF neuron for a large scale neuromorphic system", 《IEEE ESSDERC 2016 - 46TH EUROPEAN SOLID-STATE DEVICE RESEARCH CONFERENCE - LAUSANNE, SWITZERLAND》 * |
| ANDREW NERE ET.AL: "Bridging the semantic gap: Emulating biological neuronal behaviors with simple digital neurons", 《 2013 IEEE 19TH INTERNATIONAL SYMPOSIUM ON HIGH PERFORMANCE COMPUTER ARCHITECTURE (HPCA)》 * |
| 王向文: "基于脉冲序列内积的脉冲神经网络监督学习研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018133570A1 (en) * | 2017-01-20 | 2018-07-26 | 清华大学 | Self-adaptive threshold neuron information processing method, self-adaptive leakage value neuron information processing method and system, and computer device and readable storage medium |
| CN107609640A (en) * | 2017-10-01 | 2018-01-19 | 胡明建 | A kind of threshold values selects the design method of end graded potential formula artificial neuron |
| WO2022063202A1 (en) * | 2020-09-25 | 2022-03-31 | 北京灵汐科技有限公司 | Text classification method, apparatus, device, and storage medium |
| CN112734012A (en) * | 2021-01-07 | 2021-04-30 | 北京灵汐科技有限公司 | Impulse neural network training method, data processing method, electronic device, and medium |
| US11941526B2 (en) | 2021-01-07 | 2024-03-26 | Lynxi Technologies Co., Ltd. | Methods, electronic devices, and computer-readable media for training, and processing data through, a spiking neuron network |
| CN112734012B (en) * | 2021-01-07 | 2024-03-05 | 北京灵汐科技有限公司 | Impulse neural network training method, data processing method, electronic equipment and medium |
| CN112766469A (en) * | 2021-01-22 | 2021-05-07 | 北京灵汐科技有限公司 | Method and device for generating pulse signal |
| WO2022156759A1 (en) * | 2021-01-22 | 2022-07-28 | 北京灵汐科技有限公司 | Method and apparatus for generating pulse signal, and electronic device and readable storage medium |
| CN112766469B (en) * | 2021-01-22 | 2024-01-30 | 北京灵汐科技有限公司 | Method and device for generating pulse signal |
| CN112949833A (en) * | 2021-03-26 | 2021-06-11 | 北京航空航天大学 | Probability calculation neuron calculation unit and construction method |
| CN113449864B (en) * | 2021-07-21 | 2023-08-25 | 北京大学 | Feedback type impulse neural network model training method for image data classification |
| CN113449864A (en) * | 2021-07-21 | 2021-09-28 | 北京大学 | Feedback type pulse neural network model training method for image data classification |
| CN114861838A (en) * | 2022-07-06 | 2022-08-05 | 武汉理工大学 | Intelligent classification method for pulsatile neural brains based on neuron complex dynamics |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106875005B (en) | 2019-09-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106875005A (en) | Adaptive threshold neuronal messages processing method and system | |
| CN106875003A (en) | Self adaptation leakage value neuronal messages processing method and system | |
| CN106875004A (en) | Composite mode neuronal messages processing method and system | |
| CN110555523B (en) | Short-range tracking method and system based on impulse neural network | |
| CN110210563A (en) | The study of pattern pulse data space time information and recognition methods based on Spike cube SNN | |
| CN107103754A (en) | A kind of road traffic condition Forecasting Methodology and system | |
| CN106845633A (en) | Neutral net information conversion method and system | |
| CN109816111A (en) | Reading understands model training method and device | |
| CN109615015A (en) | A kind of data preprocessing method based on block chain intelligence contract and machine learning | |
| CN108121945A (en) | A kind of multi-target detection tracking, electronic equipment and storage medium | |
| CN106845632A (en) | Impulsive neural networks information is converted to the method and system of artificial neural network information | |
| CN114692681B (en) | SCNN-based distributed optical fiber vibration and acoustic wave sensing signal identification method | |
| CN110472154A (en) | A kind of resource supplying method, apparatus, electronic equipment and readable storage medium storing program for executing | |
| CN108683526B (en) | Method for identifying competitive MAC protocol | |
| CN106991666A (en) | A kind of disease geo-radar image recognition methods suitable for many size pictorial informations | |
| WO2018133570A1 (en) | Self-adaptive threshold neuron information processing method, self-adaptive leakage value neuron information processing method and system, and computer device and readable storage medium | |
| CN110288878A (en) | Adaptive learning method and device | |
| CN106875006A (en) | Artificial neuron metamessage is converted to the method and system of spiking neuron information | |
| CN104077524B (en) | Training method and viruses indentification method and device for viruses indentification | |
| CN108091194A (en) | Intelligent clearance examination management method | |
| CN108229640A (en) | The method, apparatus and robot of emotion expression service | |
| CN113553918A (en) | Machine-made invoice character recognition method based on pulse active learning | |
| CN106815638A (en) | The neuronal messages processing method and system that input weight is expanded | |
| CN116743468A (en) | Dynamic attack path generation method based on reinforcement learning | |
| CN116843069A (en) | Commuting flow estimation method and system based on crowd activity intensity characteristics |
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 |