CN111275161A - Competitive neural network framework based on DNA strand displacement - Google Patents

Competitive neural network framework based on DNA strand displacement Download PDF

Info

Publication number
CN111275161A
CN111275161A CN202010104480.1A CN202010104480A CN111275161A CN 111275161 A CN111275161 A CN 111275161A CN 202010104480 A CN202010104480 A CN 202010104480A CN 111275161 A CN111275161 A CN 111275161A
Authority
CN
China
Prior art keywords
neuron
chain
logic
neural network
strand displacement
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
Application number
CN202010104480.1A
Other languages
Chinese (zh)
Other versions
CN111275161B (en
Inventor
王宾
李亚
吕卉
张强
魏小鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Dalian University
Original Assignee
Dalian University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Dalian University filed Critical Dalian University
Priority to CN202010104480.1A priority Critical patent/CN111275161B/en
Publication of CN111275161A publication Critical patent/CN111275161A/en
Application granted granted Critical
Publication of CN111275161B publication Critical patent/CN111275161B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • 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
    • G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Neurology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)

Abstract

A competitive neural network framework based on DNA strand displacement belongs to the field of biological computation. The method comprises the steps of firstly, calculating input of AND, NAND AND OR logic values according to molecular logic to set DNA signal concentration, respectively constructing each logic gate through two neurons to form a small neural network, cascading the molecular logic gates realized through a competitive neural network to realize a molecular logic device XOR AND a half adder logic circuit, AND constructing a three-person voter. The molecular logic calculation and the competitive neural network are combined together, the competitive neural network based on DNA chain replacement is realized, the use of chains is reduced, and a more stable and visual simulation result can be obtained by adjusting the initial signal concentration of each module DNA chain.

Description

Competitive neural network framework based on DNA strand displacement
Technical Field
The invention relates to the field of biological computation, in particular to a competitive neural network framework based on DNA strand displacement.
Background
DNA is a biological information carrier, according to Watson-Crick base complementary pairing principle, the DNA can form a double-helix structure besides a single-strand form, and DNA strand displacement is used as an enzyme-free automatic DNA calculation technology, and can realize a complex model to solve various problems. In recent years, DNA strand displacement has become a research hotspot, and more models are used to solve various problems, such as digital logic circuits, feedback control circuits, neural networks, and the like.
The application of DNA strand displacement in a logic circuit and a neural network becomes a research hotspot, a xylonite team constructs a Hopfield neural network with a feedback function by using double-track logic and through a DNA strand displacement reaction, and a game of 'heart reading' is realized by using four neurons; in 2013, professor of the university of tokyo in japan proposes that a competitive neural network is used for realizing prediction of scientist problems, and only 23 DNA chains are used for successfully constructing, so that a DNA circuit is reduced, the effect of general eating of winner is simulated, and the fact that DNA is an exquisite substrate is proved; in 2014, biosensor proposed by professor dawn in the dawn of dawn dragon, etc. designed intelligent DNA molecular system, which is formed by connecting some specific DNA neurons in series, and can automatically perform logic calculation, including and, or logic gate; in 2018, a winner who implemented by Cherry et al eat DNA neural networks, smoothly classifies test patterns by using seesaw DNA gate motifs studied by Winfree, and proves that the patterns based on the DNA neural networks can identify 9 handwritten numbers for identifying 9 patterns; from this point of view, the neural network realized by DNA strand displacement has huge development space.
Disclosure of Invention
The invention provides a competitive neural network framework based on DNA strand displacement so as to obtain a more stable and visual simulation result.
In order to achieve the purpose, the invention adopts the technical scheme that: a competitive neural network framework based on DNA strand displacement, comprising the following steps:
s1: setting the concentration of input DNA signals according to the input molecular logic value, randomly initializing the weight, and enabling the neuron to perform strand displacement reaction to obtain an output strand, wherein the reaction process of the neuron is as follows:
Di×Ma+Gate→Pi+waste (1)
Di×Mb+Gate→Qi+waste (2)
the formula (1) is the process of the strand displacement reaction of the input single-chain Ma and the weight Gate, and the formula (2) is the process of the strand displacement reaction of the input single-chain Mb and the weight Gate; wherein Di is the input molecular logic value, when a single-chain Ma or a single-chain Mb exists, the corresponding Di is the logic value '1', and when the single-chain Ma or the single-chain Mb does not exist, the corresponding Di is the logic value '0'; waste is the waste chain generated; pi, Qi are the generated output chains;
s2: when the neuron is used for realizing the calculation of the AND in the molecular logic, the generated output chains Pi AND Qi are accumulated through the neuron reaction AND the weight sum is obtained, AND the neuron reaction process is as follows:
Pi+Qi→Zj+waste (3)
wherein Zj is the sum of the weights;
s3: adding an annihilation chain Anni and a Help chain Help in the neuron reaction to realize the calculation of NAND and OR by molecular logic, wherein the neuron reaction process is as follows:
Zj+Anni+Help→waste(4)
with the proviso that
Figure BDA0002388059380000031
S4: constructing a three-person voter, wherein the criteria are as follows:
Figure BDA0002388059380000032
therein netnNet being the neuron with the largest output valuejOther neurons; the output On of the winning neuron n is 1, and the output On of the other neurons n is 0.
Further, the step S3 further includes: AND combining AND cascading the AND, NAND AND OR of the molecular logic calculation constructed by the neurons to realize the XOR of the molecular logic devices.
Further, the method also comprises the step of constructing a half adder logic circuit based on the neuron structure, and specifically comprises the following steps: AND performing molecular logic AND calculation implemented in a neuron form on the original two inputs to obtain the output of a sum bit S, AND then performing XOR calculation to obtain a carry C of the half adder.
The invention has the beneficial effects that: the molecular logic calculation and the competitive neural network are combined together, the competitive neural network based on DNA chain replacement is realized, the use of chains is reduced, and a more stable and visual simulation result can be obtained by adjusting the initial signal concentration of each module DNA chain.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a schematic diagram of a process of constructing a molecular logic computation AND based on two neurons according to the present invention;
FIG. 3 is a simulation diagram of the present invention for constructing the AND of molecular logic computation based on two neurons;
FIG. 4 is a Carnot diagram of a three person voter according to the invention;
FIG. 5 is a simulation diagram of the representation of the three-person voter of the present invention passing through;
FIG. 6 is a simulation diagram showing the failure of the three-person voter of the present invention.
Detailed Description
A competitive neural network framework based on DNA strand displacement, comprising the following steps:
s1: setting the concentration of input DNA signals according to the input molecular logic value, randomly initializing the weight, and enabling the neuron to perform strand displacement reaction to obtain an output strand, wherein the reaction process of the neuron is as follows:
Di×Ma+Gate→Pi+waste (1)
Di×Mb+Gate→Qi+waste (2)
formula (1) is a process of performing a strand displacement reaction on the input single chain Ma and the weight Gate, formula (2) is a process of performing a strand displacement reaction on the input single chain Mb and the weight Gate, the weight Gate gates have the same structure, and random initialization of weights enables each Gate to generate different reactions on each group of input; wherein Di is the input molecular logic value, when a single-chain Ma or a single-chain Mb exists, the corresponding Di is the logic value '1', and when the single-chain Ma or the single-chain Mb does not exist, the corresponding Di is the logic value '0'; waste is the waste chain generated; pi, Qi are the generated output chains; in the embodiment, the concentrations of AND gate AND OR annihilation gate are 150 AND 50 respectively, the NAND adopts two annihilation gates, the concentrations are 150 AND 200 respectively, when inputting, if the input logic value is '1', the input is specific concentration input, otherwise, the input concentration value is 0;
s2: when the neuron is used for realizing the calculation of the AND in the molecular logic, the generated output chains Pi AND Qi are accumulated through the neuron reaction AND the weight sum is obtained, AND the neuron reaction process is as follows:
Pi+Qi→Zj+waste (3)
wherein Zj is the sum of the weights;
s3: annihilation chain Anni and Help chain Help are added in the neuron reaction to counteract the generated chain, thereby realizing the calculation of NAND and OR by molecular logic, and the neuron reaction process is as follows:
Zj+Anni+Help→waste (4)
with the proviso that
Figure BDA0002388059380000051
Wherein the annihilation chain Anni has a limiting effect on the regeneration of the output chain, while the addition of the annihilation chain Anni promotes competition, the annihilation chain acts as a lateral inhibition, which is some inhibition of other neurons of the output layer, each neuron tends to inhibit the neurons of the same layer as the other neurons, only one output neuron is active at a time during the reaction to determine the winning party.
Combining AND cascading AND, NAND AND OR of molecular logic calculation constructed by neurons, wherein the first layer is AND AND NAND, the second layer is OR, AND the output result of the first layer is used as the input of the second layer, so that the problem of linear inseparable XOR can be solved by using a competitive neural network, AND the XOR of molecular logic devices can be realized;
constructing a logic circuit of a half adder based on a neuron structure, which specifically comprises the following steps: performing molecular logic AND calculation implemented in a neuron form on the two original inputs to obtain the output of a sum bit S, AND then performing XOR calculation to obtain a carry C of the half adder;
s4: constructing a three-person voter, wherein the criteria are as follows:
Figure BDA0002388059380000052
therein netnNet being the neuron with the largest output valuejOther neurons; the output On of the winning neuron n is 1, and the output On of the other neurons n is 0;
it follows the rules of "minority subject to majority" such as ABC is the opinion of three persons as shown in fig. 4, 5 and 6, the keys agree to a logic "1" and mean different to a logic "0", L indicates the result of the voter, events pass through a logic "1", events fail to pass through a logic "0", the rule of the program applying the three-person voter is that the player promotes (promotion is indicated by 1) is indicated only when two persons and more than two persons support (support representative input is 1); otherwise, eliminating the player (the output is 0), and the logic formula is as follows:
Figure BDA0002388059380000061
the above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be able to cover the technical solutions and the inventive concepts of the present invention within the technical scope of the present invention.

Claims (3)

1. A competitive neural network framework based on DNA strand displacement, which is characterized by comprising the following steps:
s1: setting the concentration of input DNA signals according to the input molecular logic value, randomly initializing the weight, and enabling the neuron to perform strand displacement reaction to obtain an output strand, wherein the reaction process of the neuron is as follows:
Di×Ma+Gate→Pi+waste (1)
Di×Mb+Gate→Qi+waste (2)
the formula (1) is the process of the strand displacement reaction of the input single-chain Ma and the weight Gate, and the formula (2) is the process of the strand displacement reaction of the input single-chain Mb and the weight Gate; wherein Di is the input molecular logic value, when a single-chain Ma or a single-chain Mb exists, the corresponding Di is the logic value '1', and when the single-chain Ma or the single-chain Mb does not exist, the corresponding Di is the logic value '0'; waste is the waste chain generated; pi, Qi are the generated output chains;
s2: when the neuron is used for realizing the calculation of the AND in the molecular logic, the generated output chains Pi AND Qi are accumulated through the neuron reaction AND the weight sum is obtained, AND the neuron reaction process is as follows:
Pi+Qi→Zj+waste (3)
wherein Zj is the sum of the weights;
s3: adding an annihilation chain Anni and a Help chain Help in the neuron reaction to realize the calculation of NAND and OR by molecular logic, wherein the neuron reaction process is as follows:
Zj+Anni+Help→waste (4)
with the proviso that
Figure FDA0002388059370000011
S4: constructing a three-person voter, wherein the criteria are as follows:
Figure FDA0002388059370000021
therein netnNet being the neuron with the largest output valuejOther neurons; the output On of the winning neuron n is 1, and the output On of the other neurons n is 0.
2. The competitive neural network framework based on DNA strand displacement according to claim 1, wherein the step S3 further comprises: AND combining AND cascading the AND, NAND AND OR of the molecular logic calculation constructed by the neurons to realize the XOR of the molecular logic devices.
3. The competitive neural network framework based on DNA strand displacement according to claim 2, further comprising a half adder logic circuit based on a neuron structure, specifically: AND performing molecular logic AND calculation implemented in a neuron form on the two original inputs Di to obtain the output of a sum bit S, AND then performing XOR calculation to obtain a carry C of the half adder.
CN202010104480.1A 2020-02-20 2020-02-20 Competing neural network framework based on DNA strand displacement Active CN111275161B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010104480.1A CN111275161B (en) 2020-02-20 2020-02-20 Competing neural network framework based on DNA strand displacement

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010104480.1A CN111275161B (en) 2020-02-20 2020-02-20 Competing neural network framework based on DNA strand displacement

Publications (2)

Publication Number Publication Date
CN111275161A true CN111275161A (en) 2020-06-12
CN111275161B CN111275161B (en) 2023-05-30

Family

ID=71003572

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010104480.1A Active CN111275161B (en) 2020-02-20 2020-02-20 Competing neural network framework based on DNA strand displacement

Country Status (1)

Country Link
CN (1) CN111275161B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111832726A (en) * 2020-07-30 2020-10-27 郑州轻工业大学 Implementation method of three-dimensional chaotic oscillation system PI control based on DNA strand displacement
CN112331258A (en) * 2020-11-06 2021-02-05 大连大学 Artificial neuron calculation model construction method based on DNA cage-like structure
CN112348178A (en) * 2020-11-06 2021-02-09 大连大学 Artificial neural network calculation model construction method based on DNA strand displacement
CN113762513A (en) * 2021-09-09 2021-12-07 沈阳航空航天大学 DNA neuron learning method based on DNA strand displacement

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107480106A (en) * 2017-07-06 2017-12-15 大连大学 The construction method of dimensional probability distribution model based on strand displacement
CN108710780A (en) * 2018-04-04 2018-10-26 大连大学 A kind of DNA net structure methods regulating and controlling E6 ribozyme functions based on strand displacement
CN109797181A (en) * 2019-01-28 2019-05-24 沈阳航空航天大学 A kind of long-range DNA chain displacement method of magnesium ion induction E6-type DNA ribozyme allosteric

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107480106A (en) * 2017-07-06 2017-12-15 大连大学 The construction method of dimensional probability distribution model based on strand displacement
CN108710780A (en) * 2018-04-04 2018-10-26 大连大学 A kind of DNA net structure methods regulating and controlling E6 ribozyme functions based on strand displacement
CN109797181A (en) * 2019-01-28 2019-05-24 沈阳航空航天大学 A kind of long-range DNA chain displacement method of magnesium ion induction E6-type DNA ribozyme allosteric

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
YANFENG WANG 等: "Molecular Logic Gates Based on Localized DNA Strand Displacement", JOURNAL OF COMPUTATIONAL AND THEORETICAL NANOSCIENCE *
卢晓建;张晓康;王宾;张强;魏小鹏;: "基于Seesaw门的双向联想记忆神经网络", 广州大学学报(自然科学版) *

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111832726A (en) * 2020-07-30 2020-10-27 郑州轻工业大学 Implementation method of three-dimensional chaotic oscillation system PI control based on DNA strand displacement
CN111832726B (en) * 2020-07-30 2022-02-15 郑州轻工业大学 Implementation method of three-dimensional chaotic oscillation system PI control based on DNA strand displacement
CN112331258A (en) * 2020-11-06 2021-02-05 大连大学 Artificial neuron calculation model construction method based on DNA cage-like structure
CN112348178A (en) * 2020-11-06 2021-02-09 大连大学 Artificial neural network calculation model construction method based on DNA strand displacement
CN112331258B (en) * 2020-11-06 2024-02-23 大连大学 Artificial neuron calculation model construction method based on DNA cage structure
CN112348178B (en) * 2020-11-06 2024-03-29 大连大学 Artificial neural network calculation model construction method based on DNA strand displacement
CN113762513A (en) * 2021-09-09 2021-12-07 沈阳航空航天大学 DNA neuron learning method based on DNA strand displacement
CN113762513B (en) * 2021-09-09 2023-09-29 沈阳航空航天大学 DNA neuron learning method based on DNA strand displacement

Also Published As

Publication number Publication date
CN111275161B (en) 2023-05-30

Similar Documents

Publication Publication Date Title
CN111275161A (en) Competitive neural network framework based on DNA strand displacement
Chandra Competition and collaboration in cooperative coevolution of Elman recurrent neural networks for time-series prediction
Fiesler et al. Handbook of neural computation
CN107437096B (en) Image classification method based on parameter efficient depth residual error network model
Yao et al. Evolving artificial neural network ensembles
CN106485317A (en) A kind of neutral net accelerator and the implementation method of neural network model
CN100390774C (en) Plausible neural network with supervised and unsupervised cluster analysis
Fausett Fundamentals of neural networks: architectures, algorithms and applications
Chen et al. PERSON—personalized expert recommendation system for optimized nutrition
Plebe et al. The unbearable shallow understanding of deep learning
CN106164939A (en) Spike is provided the training in degree of depth confidence network (DBN), identification and generates
CN108647251A (en) The recommendation sort method of conjunctive model is recycled based on wide depth door
JPH06509669A (en) improved neural network
CN103930907B (en) A kind of method and apparatus of nerve setup memory transfer
Sarigül et al. Performance comparison of different momentum techniques on deep reinforcement learning
CN112634019A (en) Default probability prediction method for optimizing grey neural network based on bacterial foraging algorithm
Golosio et al. Thalamo-cortical spiking model of incremental learning combining perception, context and NREM-sleep
CN110544511A (en) four-input factorial addition operation molecular circuit design method based on DNA strand displacement
CN108229640A (en) The method, apparatus and robot of emotion expression service
Belciug et al. Data mining-based intelligent decision support systems
Videgaín et al. Performance study of minimax and reinforcement learning agents playing the turn-based game iwoki
CN111617479B (en) Acceleration method and system of game artificial intelligence system
Khan et al. In search of intelligence: evolving a developmental neuron capable of learning
CN110354501A (en) Behavior prediction method, apparatus and electronic equipment
Ünal et al. Neural Logic Circuits: An evolutionary neural architecture that can learn and generalize

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