CN113723602A - Nerve synapse scheme of optical fiber structure - Google Patents
Nerve synapse scheme of optical fiber structure Download PDFInfo
- Publication number
- CN113723602A CN113723602A CN202111021706.2A CN202111021706A CN113723602A CN 113723602 A CN113723602 A CN 113723602A CN 202111021706 A CN202111021706 A CN 202111021706A CN 113723602 A CN113723602 A CN 113723602A
- Authority
- CN
- China
- Prior art keywords
- pulse
- synaptic
- optical fiber
- change material
- synapse
- 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.)
- Pending
Links
- 210000000225 synapse Anatomy 0.000 title claims abstract description 46
- 239000013307 optical fiber Substances 0.000 title claims abstract description 40
- 210000005036 nerve Anatomy 0.000 title abstract description 9
- 239000012782 phase change material Substances 0.000 claims abstract description 34
- 230000000946 synaptic effect Effects 0.000 claims abstract description 21
- 230000003287 optical effect Effects 0.000 claims abstract description 14
- 210000002569 neuron Anatomy 0.000 claims abstract description 9
- 239000000835 fiber Substances 0.000 claims abstract description 7
- 230000001242 postsynaptic effect Effects 0.000 claims description 20
- 230000003518 presynaptic effect Effects 0.000 claims description 12
- 238000002310 reflectometry Methods 0.000 claims description 8
- 238000002425 crystallisation Methods 0.000 claims description 7
- 230000008025 crystallization Effects 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 5
- 238000001755 magnetron sputter deposition Methods 0.000 claims description 4
- -1 chalcogenide compound Chemical class 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 claims description 3
- 238000005265 energy consumption Methods 0.000 claims description 2
- 238000002347 injection Methods 0.000 claims description 2
- 239000007924 injection Substances 0.000 claims description 2
- 239000010409 thin film Substances 0.000 claims description 2
- 230000001590 oxidative effect Effects 0.000 claims 1
- 238000013528 artificial neural network Methods 0.000 abstract description 2
- 238000011161 development Methods 0.000 abstract description 2
- 210000004556 brain Anatomy 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000000034 method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 210000004242 electrical synapse Anatomy 0.000 description 2
- 230000001105 regulatory effect Effects 0.000 description 2
- 230000002441 reversible effect Effects 0.000 description 2
- 230000003977 synaptic function Effects 0.000 description 2
- 102000005591 NIMA-Interacting Peptidylprolyl Isomerase Human genes 0.000 description 1
- 108010059419 NIMA-Interacting Peptidylprolyl Isomerase Proteins 0.000 description 1
- 229910052581 Si3N4 Inorganic materials 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013529 biological neural network Methods 0.000 description 1
- 238000005253 cladding Methods 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000009123 feedback regulation Effects 0.000 description 1
- 239000010408 film Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 230000003278 mimic effect Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000000377 silicon dioxide Substances 0.000 description 1
- HQVNEWCFYHHQES-UHFFFAOYSA-N silicon nitride Chemical compound N12[Si]34N5[Si]62N3[Si]51N64 HQVNEWCFYHHQES-UHFFFAOYSA-N 0.000 description 1
- 239000000243 solution Substances 0.000 description 1
- 238000012421 spiking Methods 0.000 description 1
- 230000000638 stimulation Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000003956 synaptic plasticity Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
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
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/067—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using optical means
-
- 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
Abstract
The invention discloses a nerve synapse scheme of an optical fiber structure. The synapse structure of a biological neuron is simulated by combining the end face of an optical fiber with a phase-change material, a pulse synapse scheme is built by using an all-fiber device, different weights of synapses are switched by adopting different pulse widths, and based on an STDP rule, pulses of the overlapped part of an outgoing light pulse and an incident light pulse exceed a threshold power and act on the optical fiber synapses together to update the weights of the outgoing light pulse and the incident light pulse. The device can realize the function of automatic regulation of synaptic weights, and all weights can be switched with each other, and the switching frequency reaches 1012. Compared with the pulse synapse based on electronic components, the all-optical pulse synapse has the advantages of interference resistance, low power consumption, simple structure, high regulation speed and the like, and is expected to provide an important direction for the development of an optical neural network.
Description
Technical Field
The invention relates to the field of optical fiber intelligent devices, in particular to a nerve synapse scheme of an optical fiber structure based on a phase change material.
Background
The neuromorphic computing is computing support of the third generation Artificial Intelligence (AI), simulates an information processing mode of a human brain, integrates storage and computing, is expected to break the constraint framework of von Neumann, and is novel computing which has autonomous learning capability and processes unstructured information in real time. The core technology of neuromorphic computation is the implementation of a "spiking neural network" (SNN), which, like a biological neural network, is formed by various types of neurons and synapses connected together. In the human brain, there are approximately 10 neurons11And synapses of about 1015Synapses are a key part of information transmission between neurons, and in the basic unit of cognitive behavior of a neuron network, synapses dominate the structure of the neuron network and are responsible for a large amount of parallelism, structural plasticity and robustness of the brain. Therefore, synaptic devices that mimic biological synaptic function and plasticity will become an important component of brain inspiring computing systems.
The Phase Change Material (PCM) has the advantages of high writing/erasing speed, large difference between the electrical and optical properties of an amorphous state and a crystalline state, good reversible cycle, good thermal stability and the like, and is a technically mature optimal selection material for realizing a synapse function. Therefore, many synapse schemes with PCM have been proposed. In electrical, for example, electrical impulses induce resistive changes in synaptic devices that, while they can perform synaptic functions, hinder their further development due to the inherent signal interference and large energy consumption of the electrical system. To avoid the electrical induced defects, Harish Bhascaran et al proposed a fully integrated all-optical synapse scheme on a silicon nitride/silica platform (Zengguang Cheng, Carlos, Wolfram H.P.Pernice, et al.on-chip photosynosine [ J ]. science. Advances,2017,3: e1700160.) that overcomes the disadvantages of electrical synapses, but its fabrication process is complex, high requirements for equipment, and difficult to apply widely. The optical fiber is well coupled with light, has a simple structure and low attenuation, is similar to a neuron in structure and function, and can overcome the electrical defect, and the combination of the optical fiber and the PCM can better realize the synapse function of the neuron.
At present, optical fibers have been widely used in the fields of optical fiber communication and optical fiber sensing due to the characteristics of high speed, large bandwidth, good expandability, high transmission quality, low cost, strong anti-interference capability, and the like. The nerve synapse scheme based on the optical fiber structure of the phase-change material has the advantages of simple structure, low power consumption, fast operation and weight feedback regulation, and can realize the STDP learning rule.
Disclosure of Invention
The invention aims to provide a nerve synapse scheme based on an optical fiber structure of a phase change material, which realizes an STDP (standard deviation test) learning rule of synapses by inputting single pulses with fixed power and different pulse widths to adjust the weight of the synapses.
The neurosynaptic scheme of the optical fiber structure is realized by the following steps:
a neurosynaptic scheme of a fiber optic structure, comprising: pulsed light source (1), phase modulator (2), 50:50 coupler (3), circulator (4), synaptic structure (5), 90: 10 beam splitter (6), photodetector (7), delay line (8), optical amplifier (9). The pulse light source (1) is used for generating pulse light meeting requirements, such as certain wavelength, power, pulse width, frequency and the like; the phase modulator (2) is used for eliminating the interference of presynaptic pulses and postsynaptic pulses; 50: the coupler (3) is used for superposing presynaptic pulses and processed postsynaptic pulses; the circulator (4) is used for transmitting the pulse from a1 to a2 to the synaptic structure (5) and transmitting the reflected pulse from a2 to a3 to the postsynaptic structure; 90: the 10 beam splitter (6) is used for splitting the post-synaptic pulse into two beams of 90% and 10%, wherein 90% of the two beams flow into the delay line (8) and 10% of the two beams flow into the photodetector (7); the photoelectric detector (7) is used for detecting the shape of the postsynaptic pulse and determining the pulse width at the moment; the delay line (8) is a tunable delay line, which is used to delay the post-synaptic pulse to different degrees, and then to amplify the power of the post-synaptic pulse to the same power as the pre-synaptic pulse through an optical amplifier (9).
Further, the synapse structure (5) is composed of an optical fiber (5-1), a phase change material (5-2), and a cladding layer (5-3), characterized in that: the phase-change material (5-2) is directly covered on the end face of the optical fiber (5-1), and the covering layer (5-3) is directly covered above the phase-change material (5-2). The optical fiber (5-1) can be a single-mode optical fiber or a multi-mode optical fiber, the phase-change material (5-2) can be selected from chalcogenide compounds and consists of any two or more of three elements of Ge, Sb and Te, the optical property of the chalcogenide compounds can be adjusted by incident light pulses, and the covering layer (5-3) can be selected from an Au thin film to prevent the phase-change material from being oxidized and improve the reflectivity of a synapse structure. The phase-change material (5-2) and the covering layer (5-3) can be plated on the end face of the optical fiber (5-1) by means of radio frequency magnetron sputtering and direct current magnetron sputtering respectively. Wherein, the optical property of the phase-change material (5-2) is different according to the crystallization degree, and the higher the crystallization degree is, the higher the reflectivity of the communication waveband light is. The synaptic weight is regulated by the crystallization degree of the phase-change material (5-2), which is regulated by the pulse light of different pulse widths with input fixed power. The light injection and light collection of the optical fiber synapse structure (5) adopt an optical fiber circulator, wherein the received light pulse is the light pulse reflected by the phase change material (5-2) and the covering layer (5-3). The synapse structure (5) can realize the adjustment and reading of synapse weight by different optical signals, and the adjustment frequency of the synapse weight reaches 1012。
Further, by 50:50 the pre-synaptic and post-synaptic pulses of the coupler (3) can exceed the threshold for adjusting the weight of the synaptic structure (5) only when the pulse widths of the two overlap, and the non-overlapping part does not exceed the threshold and does not cause the weight to change.
The STDP learning rule depends on the order and spacing of external signal pulses that reach the synapse. Synaptic weights may change from increasing (inhibiting) to decreasing (increasing) when the time sequence of pulses to the synapse changes. This phenomenon is called asymmetric STDP in the brain, and conversely, symmetric STDP; when the time interval of two rows of pulses reaching synapses changes, the synapse weight changes exponentially; the STDP is considered as the most important learning mechanism in the Hebbian theory and plays an important role in information coding, learning and memorizing, and the learning formula is as follows:
in the formula, Δ W represents the change of the weight of the nerve synapse, Δ t represents the relative time difference of arrival of the pre-synaptic and post-synaptic impulse signals, τ and +/-represent time constants and exponential factors, and a is a constant.
In the invention, different pulse widths with different overlapping degrees can be realized by adjusting different delay time delta t of the delay line (8), so that synapses obtain different weights, and the STDP learning rule is realized.
The fully integrated all-photon synapse device is constructed by combining the phase-change material and the optical fiber, the phase-change material has the advantages of high writing/erasing speed, good reversible cycle, good thermal stability, large difference between the electronic and optical performances of an amorphous state and a crystalline state and the like, and is a technically mature optimal selection material for realizing synapse function; the method has the advantages that the phase-change material has different responses to different pulses, the crystallinity of the phase-change material is changed under the stimulation of different pulse widths, the performances such as reflectivity, absorptivity and refractive index of the phase-change material are influenced, the long-term stability is kept, the multi-level reflection state of a synapse device is realized under the Pulse Width Modulation (PWM), and the pulse time sequence dependence plasticity (STDP) is realized by using an adjustable delay circuit.
Advantageous effects
Compared with the prior art, the invention has the following beneficial effects:
1. the invention provides a mode of combining a phase change material and optical fibers, constructs an optical fiber nerve synapse structure, gives optical fibers 'wisdom', realizes an STDP learning rule of synapse, and develops a new direction for simulating biological synapse;
2. compared with the electrical synapse, the all-optical fiber nerve synapse based on the phase-change material has the advantages of simple natural structure, high efficiency and low power consumption.
Drawings
1. FIG. 1 is a schematic diagram of a neurosynaptic scheme for an optical fiber structure;
2. FIG. 2 is a schematic diagram of a synapse structure;
3. FIG. 3 is a schematic diagram of Pulse Width Modulation (PWM) in the present invention;
4. FIG. 4 is an exponential plot of synaptic weight as a function of time delay (Δ t) for pre-and post-synaptic pulses in accordance with the present invention;
Detailed Description
The technical process of the present invention will be described in further detail by the embodiments with reference to the accompanying drawings, and the described embodiments are merely illustrative of the present invention and are not intended to limit the present invention.
As shown in fig. 1 and 2, in the example, the wavelength of the pulsed light is 1550nm, the power is 95mW, and the pulse width is 70 ns; the optical fiber (5-1) of the synapse structure is single mode optical fiber, and the phase change material (5-2) is Ge2Sb2Te5(GST), and Au film is selected as the covering layer (5-3). The amorphous state of GST has low reflectivity, the crystalline state has high reflectivity, GST is changed into FCC structure when heated to about 420K by external light/electric pulse, and the GST is presented as crystalline state, and the GST can return to the amorphous state when the material is heated to the melting temperature (about 890K) by the pulse with higher energy and cooled rapidly, therefore, the GST can be switched by the pulses with different pulse widths with different energy to obtain different crystalline states of GST; the reflectivity of GST is different for different crystalline states, and the synaptic weights are different for different phases.
Defining the pulse output by the pulse light source (1) as In1, the post-synaptic pulse as In2, the pre-synaptic pulse output by the coupler (3) as Out, and the corresponding pulse power as PIn1、PIn2And Pout. Defining GST threshold switching power as Pth. The pulse output by the pulse light source (1) is connected to the phase modulator (2) to eliminate the interference of In1 and In2, 50% of PIN1 and 50% of PIN2Coupled together to form a presynaptic pulse from a circulator (4) a1 to a2, the reflected pulse, i.e., the post-synaptic pulse, is then passed through a circulator (4) a2 to a3, and then split into two beams by a beam splitter (6), 10% of which are transmitted to a photodetector (7) to observe the pulse shape, 90% of which are delayed by a fiber delay line (8) by a nanosecond level, and the power of which is amplified by an erbium-doped fiber optical amplifier (9) to match the power of the input pulse, such that P is PIn2＝PIn1<Pth，POut＝PIn1+PIn2When the delay time of In1 and In2 is less than the pulse width, the overlapped part of two columns of pulses is PoutAnd P isout>PthWhen the delay time of In1 and In2 is longer than the pulse width and shorter than the pulse interval, the two columns of pulses do not overlap, and the synaptic weight does not change. By adjusting the delay line (8), τ of Out gets 0 to τ as shown in fig. 3max(pulse width) and the pulse power corresponding to tau is higher than PthAnd updating the synaptic weights. The delay time given in fig. 4 is exponential with synaptic weight, satisfying the learning rule of STDP.
The above-mentioned embodiments are only a part of the embodiments, but not all embodiments, for further describing the technical solutions and advantages of the present invention in detail. It should be understood that the above-mentioned embodiments are only exemplary of the present invention, and are not intended to limit the present invention, and any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (9)
1. A neurosynaptic scheme of a fiber optic structure, comprising: pulsed light source (1), phase modulator (2), 50:50 coupler (3), circulator (4), synapse structure (5), 90: 10 beam splitter (6), photodetector (7), delay line (8), optical amplifier (9).
2. A neurosynaptic scheme according to claim 1, characterized in that the phase modulator (2) acts to eliminate interference of pre-synaptic and post-synaptic pulses; the function of the 50:50 coupler (3) is to superimpose presynaptic pulses and processed postsynaptic pulses; the circulator (4) is used for transmitting the pulse from a1 to a2 to the synaptic structure (5) and transmitting the reflected pulse from a2 to a3 to the postsynaptic structure; 90: the 10 beam splitter (6) is used for splitting the post-synaptic pulse into two beams of 90% and 10%, wherein 90% of the two beams flow into the delay line (8) and 10% of the two beams flow into the photodetector (7); the photoelectric detector (7) is used for detecting the shape of the postsynaptic pulse and determining the pulse width at the moment; the delay line (8) is a tunable delay line, which is used to delay the post-synaptic pulse to different degrees, and then to amplify the power of the post-synaptic pulse to the same power as the pre-synaptic pulse through an optical amplifier (9).
3. A neurosynaptic scheme according to claims 1 and 2, characterized in that the synapse structure (5) is composed of optical fibers (5-1), phase change material (5-2), and a cover layer (5-3), characterized in that: the phase-change material (5-2) is directly covered on the end face of the optical fiber (5-1), and the covering layer (5-3) is directly covered above the phase-change material (5-2).
4. A synaptic structure (5) according to claims 1-3, wherein the optical fiber (5-1) is a single-mode optical fiber or a multimode optical fiber, the phase change material (5-2) is a chalcogenide compound, consisting of any two or more of the three elements Ge, Sb, Te, the optical properties of which can be modulated by an incident optical pulse, and the capping layer (5-3) is an Au thin film, preventing the phase change material from oxidizing and increasing the reflectivity of the synaptic structure. The phase-change material (5-2) and the covering layer (5-3) are plated on the end face of the optical fiber (5-1) respectively in a radio frequency magnetron sputtering mode and a direct current magnetron sputtering mode. Wherein, the optical property of the phase-change material (5-2) is different according to the crystallization degree, and the higher the crystallization degree is, the higher the reflectivity of the communication waveband light is. The synaptic weight is modulated by the degree of crystallization of the phase change material (5-2). The degree of crystallization is adjusted by inputting pulsed light of different pulse widths with fixed power.
5. A neurosynaptic scheme according to any one of claims 1-4, characterized in that the light injection and light collection of the fiber optic synapse structure (5) employs a fiber optic circulator, wherein the received light pulses are light pulses reflected by the phase change material (5-2) and the cover layer (5-3).
6. A neurosynaptic scheme according to any one of claims 1-5, wherein the device can perform the adjustment and reading of synaptic weights with different light signals, and the number of times of the adjustment of synaptic weights is up to 1012。
7. The neurosynaptic scheme according to any one of claims 1-6, wherein pre-synaptic and post-synaptic pulses through a 50:50 coupler (3) can exceed a threshold for adjusting the weight of the synaptic structure (5) only if the portions of the two pulse widths that overlap, and the non-overlapping portions do not exceed the threshold and cause a change in the weight.
8. A neurosynaptic scheme according to any one of claims 1-7, characterized in that different pulse widths with different degrees of overlap can be achieved by adjusting different delay times of the delay line (8), resulting in different weights for synapses.
9. The neurosynaptic scheme according to any one of claims 1 to 8, wherein all optical fibers are connected, so that all-optical neurosynaptic is realized, the optical fiber neurosynaptic scheme is simple in structure and low in attenuation, the defects of electrical synaptic signal interference and high energy consumption are overcome, and the optical fibers have natural similarity with neurons in structure and function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111021706.2A CN113723602A (en) | 2021-09-01 | 2021-09-01 | Nerve synapse scheme of optical fiber structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111021706.2A CN113723602A (en) | 2021-09-01 | 2021-09-01 | Nerve synapse scheme of optical fiber structure |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113723602A true CN113723602A (en) | 2021-11-30 |
Family
ID=78680687
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111021706.2A Pending CN113723602A (en) | 2021-09-01 | 2021-09-01 | Nerve synapse scheme of optical fiber structure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113723602A (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5343555A (en) * | 1992-07-06 | 1994-08-30 | The Regents Of The University Of California | Artificial neuron with switched-capacitor synapses using analog storage of synaptic weights |
| US20120084241A1 (en) * | 2010-09-30 | 2012-04-05 | International Business Machines Corporation | Producing spike-timing dependent plasticity in a neuromorphic network utilizing phase change synaptic devices |
| WO2012169726A1 (en) * | 2011-06-08 | 2012-12-13 | Samsung Electronics Co., Ltd. | Synapse for function cell of spike timing dependent plasticity (stdp), function cell of stdp, and neuromorphic circuit using function cell of stdp |
| CN103078054A (en) * | 2013-01-04 | 2013-05-01 | 华中科技大学 | Unit, device and method for simulating biological neuron and neuronal synapsis |
| CN103282919A (en) * | 2010-12-30 | 2013-09-04 | 国际商业机器公司 | Electronic synapses for reinforcement learning |
| CN107092959A (en) * | 2017-04-07 | 2017-08-25 | 武汉大学 | Hardware friendly impulsive neural networks model based on STDP unsupervised-learning algorithms |
| CN107634140A (en) * | 2017-09-12 | 2018-01-26 | 电子科技大学 | Light based on SiNx reads nerve synapse device architecture and preparation method thereof |
| CN108470575A (en) * | 2018-03-23 | 2018-08-31 | 北京工业大学 | The full light memory device of imitative nerve based on Ge2Sb2Te5 |
| US20190096462A1 (en) * | 2017-09-27 | 2019-03-28 | International Business Machines Corporation | One-transistor synapse cell with weight adjustment |
| CN110262090A (en) * | 2019-06-28 | 2019-09-20 | 上海理工大学 | A kind of non-volatile fiber-optical switch structure and preparation method |
| CN110794635A (en) * | 2018-08-01 | 2020-02-14 | 西安电子科技大学 | Low-power-consumption optical synapse device based on vertical cavity semiconductor optical amplifier |
| CN111142186A (en) * | 2019-12-31 | 2020-05-12 | 中国科学院半导体研究所 | Nerve synapse of waveguide structure and preparation method thereof |
-
2021
- 2021-09-01 CN CN202111021706.2A patent/CN113723602A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5343555A (en) * | 1992-07-06 | 1994-08-30 | The Regents Of The University Of California | Artificial neuron with switched-capacitor synapses using analog storage of synaptic weights |
| US20120084241A1 (en) * | 2010-09-30 | 2012-04-05 | International Business Machines Corporation | Producing spike-timing dependent plasticity in a neuromorphic network utilizing phase change synaptic devices |
| CN103282919A (en) * | 2010-12-30 | 2013-09-04 | 国际商业机器公司 | Electronic synapses for reinforcement learning |
| WO2012169726A1 (en) * | 2011-06-08 | 2012-12-13 | Samsung Electronics Co., Ltd. | Synapse for function cell of spike timing dependent plasticity (stdp), function cell of stdp, and neuromorphic circuit using function cell of stdp |
| CN103078054A (en) * | 2013-01-04 | 2013-05-01 | 华中科技大学 | Unit, device and method for simulating biological neuron and neuronal synapsis |
| CN107092959A (en) * | 2017-04-07 | 2017-08-25 | 武汉大学 | Hardware friendly impulsive neural networks model based on STDP unsupervised-learning algorithms |
| CN107634140A (en) * | 2017-09-12 | 2018-01-26 | 电子科技大学 | Light based on SiNx reads nerve synapse device architecture and preparation method thereof |
| US20190096462A1 (en) * | 2017-09-27 | 2019-03-28 | International Business Machines Corporation | One-transistor synapse cell with weight adjustment |
| CN108470575A (en) * | 2018-03-23 | 2018-08-31 | 北京工业大学 | The full light memory device of imitative nerve based on Ge2Sb2Te5 |
| CN110794635A (en) * | 2018-08-01 | 2020-02-14 | 西安电子科技大学 | Low-power-consumption optical synapse device based on vertical cavity semiconductor optical amplifier |
| CN110262090A (en) * | 2019-06-28 | 2019-09-20 | 上海理工大学 | A kind of non-volatile fiber-optical switch structure and preparation method |
| CN111142186A (en) * | 2019-12-31 | 2020-05-12 | 中国科学院半导体研究所 | Nerve synapse of waveguide structure and preparation method thereof |
Non-Patent Citations (2)
| Title |
|---|
| 陈子龙 等: "忆阻器类脑芯片与人工智能", 微纳电子与智能制造, no. 04, 15 December 2019 (2019-12-15), pages 59 - 70 * |
| 陈宏伟 等: "光子神经网络发展与挑战", 中国激光, no. 05, 31 May 2020 (2020-05-31), pages 1 - 12 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Youngblood et al. | Tunable volatility of Ge2Sb2Te5 in integrated photonics | |
| US10657440B2 (en) | Optical synapse for neuromorphic networks | |
| US10395168B2 (en) | Tunable optical neuromorphic network | |
| US20200341503A1 (en) | Optical methods and devices | |
| Toole et al. | Photonic implementation of spike-timing-dependent plasticity and learning algorithms of biological neural systems | |
| JP6784644B2 (en) | Optical signal processing circuit | |
| US10338630B2 (en) | Optical computing system | |
| US20190041796A1 (en) | Integrated optical circuit for holographic information processing | |
| CN108470575A (en) | The full light memory device of imitative nerve based on Ge2Sb2Te5 | |
| Tan et al. | Monadic Pavlovian associative learning in a backpropagation-free photonic network | |
| CN113642718B (en) | Optical fiber pulse neuron construction scheme | |
| Boshgazi et al. | Virtual reservoir computer using an optical resonator | |
| Zhou et al. | Artificial biphasic synapses based on nonvolatile phase‐change photonic memory cells | |
| CN113723602A (en) | Nerve synapse scheme of optical fiber structure | |
| CA3165560C (en) | Optical synapses | |
| USH474H (en) | Optical transversal fiber with reflective taps | |
| Vatsavai et al. | Silicon photonic microring based chip-scale accelerator for delayed feedback reservoir computing | |
| JP6829700B2 (en) | Optical signal processor | |
| CN114266346A (en) | Reconfigurable optical neural network based on phase change material and application thereof | |
| Li et al. | All-Fiber Synapse Utilizing Phase Change Materials for Information Recognition and Processing | |
| WO2021074656A1 (en) | Infroton type artificial neural network | |
| EP3987453A1 (en) | Optical devices and methods | |
| Shen et al. | Large-scale neuromorphic systems enabled by integrated photonics | |
| Bilodeau et al. | Neuromorphic Silicon Photonics | |
| Chattopadhyay et al. | Sigmoid activation function generation by photonic artificial neuron (PAN) |
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 |