CN106991475A - The apparatus and method based on mark for neutral net - Google Patents

Abstract

Disclose the apparatus and method based on mark for neutral net.Framework can be used for definition node type, definition node to node connection type, and instantiation is directed to the node instance of different node types, and/or generates the example of the connection type between these nodes.HLND forms can be used for definition node type, definition node is to node connection type, instantiation is for the node instance of different node types, the example for dynamically being identified and/or being selected network subset using mark, and/or one or more connections between these nodes are generated using such subset.In order to facilitate HLND operations and ambiguousness to eliminate, the individual element of volume (for example, node, extension, connection, I/O ports) of network can be assigned at least one uniqueness mark.The mark can be used for identifying and/or quote from corresponding network element.HLND kernels may include the interface described to element formula network.

Description

The apparatus and method based on mark for neutral net

Present patent application is that international application no is PCT/US2013/032546, and international filing date is March 15 in 2013 Day, it is entitled " to be used for the dress based on mark of neutral net into the Application No. 201380025107.5 of National Phase in China Put and method " application for a patent for invention divisional application.

The cross reference of related application

The application is entitled " the ELEMENTARY NETWORK DESCRIPTION FOR submitted for 21st in September in 2011 NEUROMORPHIC SYSTEMS (the element formula network for being used for neuron morphology system is described) " U.S. Patent application No.13/ Continued case for 239,123 part, and this application is clearly included in this by quoting.

The application is related to jointly owned entitled " the TAG-BASED APPARATUS AND submitted on January 27th, 2012 METHODS FOR NEURAL NETWORKS (apparatus and method based on mark for being used for neutral net) " United States Patent (USP) Shen Please S/N.13/XXX.XXX, jointly owned entitled " the ELEMENTARY NETWORK that are submitted for 21st in September in 2011 DESCRIPTION FOR EFFICIENT IMPLEMENTATION OF EVENT-TRIGGERED PLASTICITY RULES IN NEUROMORPHIC SYSTEMS (are used in neuron morphology system efficiently realize event-triggered plasticity rule Element formula network describe) " U.S. Patent application S/N.13/239,163, in September in 2011 submit within 21st it is jointly owned Entitled " ELEMENTARY NETWORK DESCRIPTION FOR EFFICIENT MEMORY MANAGEMENT IN NEUROMORPHIC SYSTEMS's (the element formula network description for the efficient memory management being used in neuron morphology system) " The U.S. Patent application S/N.13/239,155 and jointly owned entitled " ELEMENTARY submitted for 21st in September in 2011 NETWORK DESCRIPTION FOR EFFICIENT LINK BETWEEN NEURONAL MODELS NEUROMORPHIC SYSTEMS (the element formula network description for the efficient link being used between the neuron models in neuron morphology system) " U.S. State patent application S/N.13/239,148, the full content of above-mentioned each application is incorporated by reference thereto.

Copyright

A part includes data protected by copyright disclosed in patent document.Copyright holder does not oppose anyone to this Patent file or patent are disclosed to be replicated according to its former state in the patent document or record of patent and trademark office, but except this with It is outer to retain other all copyright rights whatsoevers.

Computer program list annex on CD-ROM

The file of this patent includes the compression with the individual read-only storage file in 46 (46) in ascii text file form The duplicate copy of disk (CD-ROM).File details are given in Table 1 below.These ascii text files are comprising expression for the disclosure The code line of the exemplary realization of computer program list.The CD-ROM and each text for including and being listed in table 1 thereon The full content of part is incorporated by reference thereto.

Table 1

Technical field

The innovation is related to efficient design and the realization of artificial neural network.

Background technology

Most existing neuron models and system are included with simple unit (being referred to as neuron (neuron)) Network, these simple units are interactively with each other via the connection of referred to as cynapse.Information processing in such neuron system can be parallel Ground is performed.

There are many specific software instruments to help the model of neuroscientist's simulated nervous system.The example of these instruments May include the senior realization that may be designed primarily to use for neuroscientist, such as NEURON, GENESIS, NEST, BRIAN, One or more of and/or other senior realizations.Such instrument may usually require that a large amount of special knowledges, it may be possible to trouble , and may require customization could reach efficient performance during the simulation performed using specific software and hardware engine (especially when requiring real-time performance, such as in autonomous robot application).

Similarly, rudimentary realization (such as assembler language, low level virtual machine (LLVM) language, Java bytecode, because of chip One or more of different instruction set, and/or other rudimentary realizations) x86, ARM can be designed to^TM, and/or other silicon cores Hardware-efficient on piece is realized.However, such realization may not be suitable for the Parallel Simulation of neuron system, this be mainly because It is not to be designed to be simulated for such parallel neuron for silicon.

Generally, existing method has substantial drawback, because they can not provide enough in design neutral net Flexibility, it is desirable to which professional knowledge, and/or the customization different because of platform could utilize specialised hardware.

Be accordingly, there are to for by the mankind can read and machine it is scrutable simple and unambiguously in the way of define net The notable demand of the general high-level network description of network framework.

The content of the invention

The present invention expires particularly by providing for apparatus and method of high-level network description of neuron morphology system etc. Sufficient the demand.

One aspect of the present invention be related to it is a kind of realize neutral net by computer implemented method.In some realizations In, the network may include multiple elements.This method may include the subset for identifying this multiple element.This method may include to the subset Individual element of volume assign mark.The appointment of given mark, which may be configured such that, can generate at least a portion for including subset member The new network element of element.

In some implementations, the mark may include the unique identification for being configured to identify one or more element of volume Symbol.

In some implementations, individual element of volume can be randomly chosen from this multiple element.The individual element of volume of the subset can be wrapped Include unit.The mark may include string (strng) identifier.

In some implementations, the mark may include alpha numeric identifier, and it can be adapted to identify the corresponding of the subset The space coordinate of each element.The subset may include multiple nodes.The alpha numeric identifier may include in this multiple node The identifier of at least one node.

In some implementations, the new network element may include connection.The connection may include it is following one or more：(i) dash forward Touch, (ii) is tied, and/or the further feature associated with connection.

Another aspect of the present invention be related to it is a kind of generated in neutral net multiple connections by computer implemented method. The neutral net may include multiple elements.In some implementations, this method may include it is following one or more：(i) perform at least The first logical expression including the first mark and the second mark, is based at least partially on the execution, (ii) identifies this multiple member Element the first subset and yield in the second subset, (iii) generate the first subset at least a portion and yield in the second subset at least a portion it Between multiple connections, and/or other operations.In some implementations, one or more element of volume of the first subset may include One mark.The individual element of volume of yield in the second subset may include the second mark.

In some implementations, the individual element of volume of the first subset and/or yield in the second subset may include the node of the network.This method It may include to assign first to mark to one or more element of volume of the first subset.

In some implementations, one or both of the first mark or the second mark can be characterized by limited life cycle.

Another aspect of the present invention is related to a kind of method that dynamic divides computerization neutral net.In some implementations, This method may include it is following one or more：(i) subset of elements of the network, (ii) are identified mark is assigned to the every of the subset One element, and/or other operations.According to some realizations, the mark and the appointment can cooperate with making it possible to using single selection behaviour Make to select one or more element of volume of the subset.

In some implementations, the network may include multiple elements.The subset may include multiple nodes of this multiple element.

In some implementations, identifying the subset and being based at least partially on to perform includes one or more keywords (bag Include AND (with), NOT (non-), OR (or) and/or other keywords) Boolean expression.

In some implementations, this method may include it is following one or more：(i) identify the network another subset of elements, (ii) by another mark be assigned to each element of another subset, (iii) enable at least a portion element in the subset with Multiple connections, and/or other operations between the element of another subset.

In some implementations, one or more of this multiple connection individual connection may include one of cynapse and knot.This is more One or more of individual connection individual connection can be based at least partially on the mark and another mark to enable.

Another aspect of the present invention is related to a kind of processing unit.In some implementations, the processing unit may include to be configured Into the non-volatile memory medium of storage multiple instruction, the plurality of instruction can be carried out according to a kind of pair of method when executed The dynamic division of neutral net.This method may include it is following one or more：(i) identify the neutral net subset of elements, (ii) mark is assigned to each element of the subset of elements, and/or other operations.The mark may include to be configured to identify one The identifier of individual or multiple element of volume.In some implementations, appointment mark, which may be configured such that to generate, includes the element The new network element of subset.

In some implementations, this method can be realized using application specific integrated circuit (ASIC) using ASIC instruction set.

In some implementations, this method may include to perform one or many that is configured to identify the subset by the processing unit The mathematic(al) representation of each and every one element of volume.The mathematic(al) representation may include Boolean calculation.

In some implementations, one or more element of volume of the subset can use random selection operation to select.

In some implementations, this method may include the mark being assigned to the new network element.Mark is assigned to the son Integrate that may be configured such that can be by the network representation as digraph.

In some implementations, this method may include the second mark being assigned to the subset.Second mark can be with foregoing tags It is different.

Further aspect of the invention is related to a kind of neuroid logic.In some implementations, the neuroid is patrolled Collect series of computation machine program step and/or the instruction that may include to perform on digital processing unit.In some implementations, the logic It may include hardware logic (for example, implementing in ASIC or FPGA).

Another aspect of the present invention is related to a kind of computer readable device.In some implementations, the device may include thereon Be stored with the storage medium of at least one computer program.The program can be configured to realize Artificial neural network when executed Network.

Another aspect of the present invention is related to a kind of system.In some implementations, the system may include there are multiple nodes therewith Associated artificial neuron (for example, spike) network, controlled device (for example, robot or prosthetic appliance), and/or other groups Part.

After the following description and the appended claims are considered referring to the drawings, the these and other objects of the disclosure, Features and characteristics and operating method and function and component combination and the organizational framework of manufacture about structural element will become more It is clear, all accompanying drawings constitute the part of this specification, wherein identical reference indicates corresponding component in the various figures.So And it is to be expressly understood that accompanying drawing is only used for explaining and describing purpose, and it is not intended as the definition of restriction of this disclosure.Such as Used in this specification and appended claims, " one ", " certain " of singulative and "the" include plural referents, remove Non- context understands that regulation is really not so.

Brief description of the drawings

Fig. 1 is to explain to include the frame for the artificial neural network that multiple nodes and node are connected according to one or more realizations Figure.

Fig. 2 is the block diagram for explaining the neurode type as network object according to one or more realizations.

Fig. 3 A are the block diagrams for explaining the node interconnection according to one or more realizations.

Fig. 3 B are the block diagrams for explaining the node interconnection including uniform dendron according to one or more realizations.

Fig. 3 C are to explain the block diagram interconnected according to the non-uniform knots of one or more realizations.

Fig. 4 is the block diagram for explaining public many chamber neurons (MCN) according to one or more realizations.

Fig. 5 is to explain the exemplary pseudo code stated according to the common node of one or more realizations.

Fig. 6 is to explain the block diagram that the common node defined according to the use common node of one or more realizations is interconnected.

Fig. 7 is to explain to include the privately owned MCN of two input interfaces and single output interface according to one or more realizations Block diagram.

Fig. 8 is the block diagram for explaining the privately owned neuron interconnections according to one or more realizations.

Fig. 9 is to explain the figure that addition is marked according to the Node subsets of one or more realizations.

Figure 10 is to explain the block diagram inherited according to the free token of one or more realizations.

Figure 11 is the block diagram for the various exemplary realizations for explaining END engines.

Figure 12 is to explain the block diagram that node establishment is carried out according to the use HLND gui interfaces of one or more realizations.

Figure 13 is to explain the block diagram that Node subsets selection is carried out according to the use HLND gui interfaces of one or more realizations.

Figure 13 A are to explain the frame that Node subsets selection is carried out according to the use HLND gui interfaces of one or more realizations Figure.

Figure 13 B are to explain the frame that Node subsets selection is carried out according to the use HLND gui interfaces of one or more realizations Figure.

Figure 14 be explain according to the use HLND gui interfaces of one or more realizations carry out node selection, mark addition and Connect the block diagram of generation.

Figure 15 is to explain to be described to carry out the frame of neural computing according to the use HLND and END of one or more realizations Figure.

Figure 16 be explain according to one or more realizations can with HLND frameworks associated with computerized device block diagram.

Figure 17 be explain according to one or more realizations can with HLND frameworks associated with data flow block diagram.

All accompanying drawings disclosed herein areThe Brain companies of copyright 2013.All rights reserved.

Embodiment

Be described in detail now with reference to accompanying drawing the realization of the disclosure there is provided accompanying drawing only as illustrative example to make this Art personnel can put into practice the disclosure.It is worth noting that, following accompanying drawing and example is not intended as the model of the disclosure Enclose and be defined in single realization, on the contrary, by with some or all described or exchanges of element for being explained or combining, Qi Tashi It is also now possible.What convenient place in office, same reference numerals, which will pass through accompanying drawing, to be used to refer to same or like portion all the time Point.

In the case where some elements of these realizations can be realized partially or completely using known tip assemblies, this will only be described Those parts necessary to for understanding the disclosure of class known tip assemblies, and the other parts of such known tip assemblies in detail retouch State and will be omitted so as not to obscure the disclosure.

In this manual, show that the realization of singular component is not construed as constituting to limit；Specifically, the disclosure is intended to Cover other realizations including multiple same components, vice versa, unless clearly stated in addition herein.

In addition, the disclosure is covered herein by equivalent known to the present and the future of the component recited in explanation.

As it is used herein, term " bus " is generally intended to indicate the institute for being used for accessing cynapse and neural metamemory There are interconnection or the communication construction of type." bus " can be optics, wireless, infrared, and/or another type of communication media.Always The definite topology of line can be for example standard " bus ", hierarchy type bus, network-on-chip, the connection of address-event-expression (AER), And/or for accessing the other types of communication topology of the different memory in such as system based on pulse.

As it is used herein, term " computer ", " computing device " and " computerized equipment " may include following one Or many persons：Personal computer (PC) and/or microcomputer are (for example, desktop computer, laptop computer, and/or other PC), mainframe computer, work station, server, personal digital assistant (PDA), handheld computer, embedded computer, can compile Journey logical device, personal communicator, tablet PC, portable navigation auxiliary, the equipment, cell phone, the intelligence that are equipped with J2ME Phone, personal integrated communicaton and/or amusement equipment, and/or instruction set can be able to carry out and any of incoming data signal is handled Miscellaneous equipment.

As it is used herein, term " computer program " or " software " may include the mankind and/or the machine of perform function Any sequence of cognizable step.This class method can with including it is following one or more programming language and/or environment present： C/C++、C#、Fortran、COBOL、MATLAB^TM, PASCAL, Python, assembler language, markup language (for example, HTML, SGML, XML, VoXML), the environment of object-oriented (for example, sharing Object Request Broker's framework (CORBA)), Java^TM(for example, J2ME, Java Beans), binary runtime environment (for example, BREW), and/or other programming languages and/or environment.

As it is used herein, term " connection ", " link ", " transmission channel ", " delay line ", " wireless " may include it is any Causal link between two or more entities (either physics or logic/virtual), the link can be realized respectively Information between entity is exchanged.

As it is used herein, term " memory " may include integrated circuit and/or be adapted to be used to store data signal Other storage devices.By non-limiting example, memory may include it is following one or more：ROM、PROM、EEPROM、 DRAM, mobile DRAM, SDRAM, DDR/2 SDRAM, EDO/FPMS, RLDRAM, SRAM, " sudden strain of a muscle " are deposited (for example, NAND/NOR), recalled Hinder device memory, PSRAM, and/or other types of memory.

As it is used herein, term " microprocessor " and " digital processing unit " are generally intended to include digital processing device. By non-limiting example, digital processing device may include it is following one or more：Digital signal processor (DSP), simplify finger Order collection computer (RISC), general (CISC) processor, microprocessor, gate array are (for example, field programmable gate array (FPGA)), the woven structure of PLD, Reentrant lines (RCF), array processor, secure microprocessor, application specific integrated circuit , and/or other digital processing devices (ASIC).Such digital processing unit can be comprised on single tenth of the twelve Earthly Branches IC tube cores, or across multiple Components distribution.

As it is used herein, term " network interface " refers to any signal, the number with component, network, and/or process According to, and/or software interface.By non-limiting example, network interface may include it is following one or more：Live wire (for example, FW400, FW800 etc.), USB (such as USB2), Ethernet (for example, 10/100,10/100/1000 (Gigabit Ethernet), 10-Gig-E (10 Gigabit Ethernet) etc.), MoCA, Coaxsys be (for example, TVnet^TM), radio-frequency tuner is (for example, with interior Or OOB, cable modem etc.), Wi-Fi (802.11), WiMAX (802.16), PAN (such as 802.15), honeycomb (example Such as, 3G, LTE/LTE-A/TD-LTE, GSM etc.), IrDA races, and/or other network interfaces.

As it is used herein, term " cynapse channel ", " connection ", " link ", " transmission channel ", " delay line " and " logical Believe channel " include any two or more entity (either reality of (wired or wireless) of physics or logic/virtual Body) between link, the link realizes that information between each entity is exchanged, and can be by influence that this information exchanges one or many Individual variable is characterized.

As it is used herein, term " Wi-Fi " include it is following one or more：Ieee standard 802.11, ieee standard 802.11 variant, the standard (for example, 802.11a/b/g/n/s/v) relevant with ieee standard 802.11, and/or other wireless Standard.

As it is used herein, term " wireless " means any wireless signal, data, communication, and/or other wirelessly connect Mouthful.By non-limiting example, wave point may include it is following one or more：Wi-Fi, bluetooth, 3G (3GPP/3GPP2), HSDPA/HSUPA, TDMA, CDMA (for example, IS-95A, WCDMA etc.), FHSS, DSSS, GSM, PAN/802.15, WiMAX (802.16), 802.20, arrowband/FDMA, OFDM, PCS/DCS, LTE/LTE-A/TD-LTE, analog cellular, CDPD, satellite system System, millimeter wave or microwave system, acoustics, infrared (that is, IrDA), and/or other wave points.

It is comprehensive to look at

The disclosure particularly provide can be configured to by it is simple and unambiguously mode come in terms of defining neuroid framework Calculation machine high-level network describes apparatus and method etc..

In some implementations, a kind of computerized device can be configured to realize that high-level network describes (HLND) kernel.Should Family can be used to use for HLND kernels, and the mankind can read and the scrutable unification of machine and expression unambiguously define neuron The form network architecture.

In some implementations, HLND forms can be used for definition node type, node to node connection type, instantiation pin Node instance to different node types, using mark dynamically to identify and/or select the subset of network, using such subset To generate the example, and/or the other information associated with node and/or mark of the connection between these nodes.

HLND forms can provide some or all of flexibilities required by computational neuroscience man and can be to building The user of limited experience provides user-friendly interface in terms of mould neuron.

In some implementations, HLND kernels may include to element formula network the interface for describing (END).END engines can by with It is set to and efficiently represents neuron system in the way of independently of hardware and/or can realize that HLND models are described to for by each Plant the seamless translation for the hardware instruction that processing module is performed.

In some implementations, HLND frameworks may include graphic user interface (GUI), and it is configured to make user especially to create Build node, selection Node subsets, selection subsets are connected using figure action via GUI, and/or execution is consistent with the disclosure Other operations etc..GUI engine can be configured to generate HLND sentences, and it may correspond to above user action, without using by oneself The further input at family.HLND frameworks can be configured to HLND sentences being converted into the figure table to network that is presented by GUI Show.HLND may include one or more assemblies, and these components include (i) and use GUI network graphic description, (ii) HLND sentences Corresponding lists, and/or other components.HLND one or more assemblies can be configured to as one man represent the phase on network Same information because the change in a kind of expression can be uniformly applied in it is other represent, thereby come reflect to the network some Or all modifications.

In some implementations, HLND can be applied to arbitrary graph structure (for example, the nerve net with arbitrarily complicated framework Network).

It there is presently provided each detailed description realized of the apparatus and method of the disclosure.Although some aspects of the disclosure can To be best understood in the context of the high-level network descriptor format for designing neutral net framework, but the disclosure not by So limit and the disclosure realization can be used for being embodied as efficiently representing with hardware independent manner other systems (for example, Biological or finance) and the instruction set of optimization.

The realization of the disclosure can be for example deployed in the hardware of neuron morphology computer system and/or software realization. In some implementations, robot system may include to implement the processor in application specific integrated circuit, and the processor can be adapted to Or be configured to use in Embedded Application (such as prosthesis apparatus).

Fig. 1 illustrates a realization of the neuroid configuration available for the disclosure.The network 100 shown in Fig. 1 is wrapped Include different types of node (node type 102,104 in Fig. 1).As described in detail below, HLND frameworks allow user to move Select to state the random subset (subset 106,108 in Fig. 1) of network node and via connection (connection 110 in Fig. 1) interconnection The node of selection subsets.Some nodes (for example, node 104_1) of network 100 can be received from more than one node (example Such as, node 102_1,102_2 in Fig. 1) input.Conversely, some nodes (such as, node 102_1,102_2) are if can be to Dry node delivering output, as explained in Fig. 1.

HLND Frame Designs are comprehensive to look at

Realized according to one or more, exemplary HLND frameworks can be configured to facilitate neutral net (such as Fig. 1 network 100) design.Some realizations can provide the ability of neuroid of the description with arbitrarily complicated property.Some realizations can be with Facilitate and predefined node and/or predefined connection type are used to network generating process.That is, multiple realities of different node types Example can be generated, is laid out, and/or be connected using multiple examples of different connection types.Some realizations can be provided to new The flexible definition of node type, so that new node type may include the realization and/or net of element formula network description (END) unit class The realization of network object (for example, node layout, connective collection, and/or these combination).The node type newly defined can be in network Used in generating process.Entitled " the ELEMENTARY NETWORK DESCRIPTION FOR that END frameworks are included more than NEUROMORPHIC SYSTEMS (the element formula network for being used for neuron morphology system is described) " U.S. Patent application S/ N.13,239,123 described in.Some realizations can provide the flexible definition to connection type.Connection type may include that END is tied The realization of class, END cynapses class, and/or other classes.In one implementation, the connection type newly defined can be in network generating process In use.Some realizations can be facilitated to some or all of network elements (including node, connection, node set, and/or other nets Network element) use common tags (or label).In some implementations, mark can be used for mark node cluster and/or connection group.Mark Note can be used for each several part for being dynamically selected network.One or more Boolean calculations, such as AND, OR, NOT, and/or other Boolean calculation can be applied to mark.Some realizations can provide using graphic user interface (GUI) to realize HLND networks Ability.The user action that individual description construction can correspond in GUI.In some implementations, the model of medium complexity can be with Built up using HLND gui interfaces without using keyboard.In some implementations, HLND GUI can use touch-screen, light pen Input equipment, and/or other input technologies are operated.Some realizations can facilitate the HLND languages for defining network anatomical structure The presentation of sentence.Define network anatomical structure may include with user can read language (Natural English) come to node and/or connect into Row layout understands in order to Non-computer Majors network designer.Some realizations can provide using HLND to generate END examples Ability.

Net definitions method

Define neutral net may include definition to create how many and/or what kind of node, how to be laid out these nodes, How these node instances (for example, Fig. 1 network topology), and/or other operation are connected.In some implementations, HLND is defined Method defines new node type and/or the connection type for these new node types including (1), and (2) define the section in network How point layout (for example, to create how many and/or what kind of node, and arranges these in the network being just created Node), (3) define how these nodes are connected to each other, and/or other operations.During neural network configuration, above step 1- 3 can individually and/or sequentially repeatedly.In some implementations, above step 1 can be skipped and can replace fixed Predefined class is used during adopted network (it, which is defined, expects node type).

In some implementations, special-purpose software bag can be configured to (i) processing and define the HLND sentences and/or (ii) of network in fact Exampleization network node and connection.This processing and/or instantiation can suffer from one or more constraints, including (i) is only defined Node type and defined connection type can be instantiated and be used during HLND net definitions, (ii) only existing section Connection between point example can be instantiated, and/or other constraints.In other words, realized, connected in definition according to one or more The connection corresponding with defined node instance only can be used during the HLND connect.In some implementations, special-purpose software bag It may include END engines, it can be configured to the END examples for generating network model, and that is such as included more than is jointly owned entitled " ELEMENTARY NETWORK DESCRIPTION FOR NEUROMORPHIC SYSTEMS (are used for neuron morphology system Element formula network describe) " U.S. Patent application S/N.13/239,123 described in.

Definition node type

The definition of node type can provide the realization instruction of node, and these realize that instruction can be configured at instruction network Reason device performs particular step according to node type during node instance.In some implementations, node definition can be further The inside of node type is specified to realize (for example, specifying the dynamic of neuron type).In one or more implementations, node definition It may include the definition of the input port and/or output port of the node.

In some implementations, node type can be defined as simple node, and wherein node definition specifies the " internal of the node Construction ".The internal structure of node may include realization and/or the neuron chamber of END units (that is, neuron) (compartment) END is realized.

In some implementations, node type can be defined as complex network object, and it can be provided on how to instantiate Instruction, the instruction, and/or other instructions on how to connect each node of predefined node type.On how to connect each section The instruction of point may include the HLND descriptions to the network with arbitrarily complicated property, is configured to specify node and/or connects example The algorithm of the details of generation, and/or other instructions.Skilled artisans will appreciate that, term " network object " can be used for retouching State any network that HLND frameworks can be used to realize.

In the description of HLND frameworks, individual nodes type may include to connect and can be defined for incoming connection in node And/or one or more interfaces of outflow connection.

It is used as the node type of END units

In some implementations, END unit classes can use node to realize that details (for example, updating rule, event rules) is come Generation.The additional detail related to END unit classes is see, for example, U.S. Patent application S/N.13,239,123.

It is used as the node type of network object

In some implementations, the definition of network object can be similar to the mode of the definition of network to configure, wherein mainly Difference is that network object is reusable.I.e., it is possible to instantiate multiple examples of a network object.Network object some or All elements (for example, unit, mark, and/or other elements) can have scope, i.e. they may have and specific model Enclose the associated limited life phase.In some implementations, network object can be configured to offer and can be used for the network object The I/O interfaces being connected with other nodes.Network object can be similar toIn model building block (see, for example,http://www.mathworks.com/products/simulink/index/html), CAD (CAD) it is soft P-cell (p cells), function/class in C Plus Plus code, and/or other programming elements in part.

In some implementations, network object can be allowed to use fixed outside (that is, outside the scope of this node type) The predefined node of justice.Therefore, father node (that is, network object) and (all) child nodes (that is, node used in network object Type) node of same type may not be included.In other words, realize that the definition of ' X ' type node can be with unreal according to some Exampleization ' X ' type node.

Referring now to Fig. 2, an exemplary realization of network object is explained and is described in detail.Network object 200 can be wrapped Include it is following one or more：Net definitions, the specification of object input/output (I/O) interface, and/or other information.

Net definitions can specify the one or more steps that object instance is generated.Net definitions can be via standard HLND nets Network defines to realize.Standard HLND net definitions may include to specify the reality of each predefined node using predefined distribution function The node instance and/or layout of number of cases amount and/or its arrangement space.Standard HLND net definitions may include connective description, Connectedness between its definable node.In some implementations, connective description can be defined and/or dashed forward using the space of node Rise.As non-limiting example, the space projection of definition node may include definition node aixs cylinder and/or dendron projection (for example, (i) dendron territory, the distribution of (ii) synaptic knob, the distribution of (iii) axon ends), definition node (for example, model neuron) Aixs cylinder how to be connected to another node (for example, model neuron) dendron, and/or definition it is associated with the space projection of node Other information.

By non-limiting example, standard HLND net definitions can be used for the specified arrangement that (i) defines predefined node； (ii) many chamber neurons (for example, the set for the predefined END units being connected is tied with predefined END) are defined；(iii) define Arbitrarily complicated network including multiple neurons；And/or define the other information associated with standard HLND net definitions. During some are realized, network may include cynapse and/or knot.

In some implementations, the definition of network may include to generate using the example of algorithm, and the algorithm is configured to Above object definition come describe network object example generation one or more steps.By non-limiting example, the algorithm can Including it is following one or more：(i) using the example generating process for predefining node type and/or the such node type of definition Algorithm；(ii) using predefined node type and/or connection type and/or the example for defining such node and/or connection type Any particular algorithms of generating process；(iii) algorithm of dendron tree is defined；And/or other algorithms.

Foregoing exemplary algorithm can utilize the multiple predefined END cell types for realizing neural chamber and/or be designed The predefined END that such chamber is connected is tied into type with the algorithm of connection therebetween into define the layout of each chamber.(referring to Such as Cuntz H., Forstner, F., Borst.A and Hausser, M. (2010) " One Rule to Grow Them All:A General Theory of Neuronal Branching and Its Practical Application.PLoS Computational Biology, 6 (8) (are allowed to all growths with a rule：The general theory of neuron branch and Its practical application, PLoS calculation biologies, 6 (8)) ", entire contents are incorporated by reference in this).I/O interfaces can be specified The input/output connection realized for network object.

The definition of connection type

Realized according to one or more, in HLND, the definition of connection type, which can be provided, necessary realizes details (example Such as, rule before event, rule after event, update regular, and/or other details) include (i) END cynapses or (ii) END to generate The connection of one or both of knot.

Instantiated nodes

In some implementations, HLND can define the rule of management and control node instance.The instruction of HLND node instanceizations can be with Being provided to deciphering afterwards, these instruct and instantiate the software kit (for example, END kernels) of appropriate node.Realized according to some, During instantiation, some or all of node types can be treated equally, whether they are from simple node (for example, the END of neuron Realize) or network object (for example, overall network description).In some implementations, it may be necessary to following information instantiating and Arrangement nodes：(i) node type to be instantiated, the example quantity of (ii) node type to be instantiated, and/or other letters Breath.

With citation form, HLND instantiation sentences can use the default definition corresponding with given node type to create N example of the node type.During instantiating, additional parameter can be used to be provided for initialization by reality with especially (i) The parameter of the node type of exampleization, (ii) sets how the node being instantiated is laid out (for example, how to assign position in space Mark), additional marking is added to new node instance by (iii), and/or performs the other behaviour associated with HLND instantiations Make.In HLND, available defined node type can be instantiated unrestrictedly in opereating specification.

In some implementations, position coordinates (that is, free token) can be assigned to what is generated during node instance Node instance.In order to realize this feature, HLND frameworks can be supported to use the node being instantiated to each to assign empty Between the predefined distribution function that marks.Such a distribution function can be configured to from given probability density function n point of sampling.By By non-limiting example, HLND sentences：

Uniform(n,boundary parameters)

It can describe to sample in spatial dimension as defined in boundary parameter (boundary parameters) independent variable The n point that uniformly (Uniform) is distributed.Similarly, HLND sentences：

Normal(n,sigma,boundary parameters)

The sampling n in as the spatial dimension specified by boundary parameter (boundary parameters) can be described individual just State (Normal) distributed point.

, can be during instantiation process optionally in addition to the uniqueness id marks that the node of individual generation can have Additional markers and these additional markers are assigned to can be used for the set for identifying new instantiated nodes.In some implementations, it is special Reservation mark (for example, " IN " (input), " OUT " (output) or other special reservations are marked) can be used for specified give birth to Into unit be network input or output interface, hence in so that these nodes are (readable and/or can from accessible outside Write).Example (example) property HLND, which is invoked in list below 1, to be shown：

A) 100 ' exc ' node types (it is assumed that ' exc ' END classes/network object presence) are created

Example1_exc_neurons=(100, ' exc ')

B) 200 ' exc ' node types are created and them are distributed using given pdf (probability-distribution function)：

Example2_exc_neurons=(200, ' exc ', _ exc_parameters_, pdf)

C) 1 node of the network object type for realizing retina (retina) is created (it is assumed that previously having realized ' retina ' network object type)：

Example3_retina=(1, ' retina ', _ retina_parameters_)

List 1.

Form/extension

In some implementations, form/extension can be used during instantiation is connected.The form described above with respect to Fig. 2 Definition specifies how the node that is instantiated projection and/or can be expanded in cyberspace.That is, form can be prolonged with definition space And scope, wherein the node being instantiated is allowed to the incoming connection of (i) " reception " from the spatial spread, and/or (ii) to The spatial spread " transmission " outflow connection.Note, according to some realizations, the addition of point spread can not change the big of the node Small and/or position.Replace, each extension can enable node " search " other nodes during the instantiation of node interconnection. In other words, extension can provide additional " view " of node, and the view can be used during the process of connecting node.

In some implementations, it only can just allow to carry out connecting node using free token in the case of each node crossover. Each node can be given tacit consent to zero extension.In some implementations, only coexistence node can be connected., can in order to which expanding node is connective To define non-zero node input (dendron) and node output (aixs cylinder) projection.In some implementations, these projections can appoint in connection What used during two nodes.For example, the output projection of a node can be connected to the input projection of another node.

In order to create extension, realized according to one or more, it may be required that some or all of in following information：(1) use In the source marking for being designated its each node for creating extension；(2) it is used for mark by the extending marking of the extension of establishment；And/or (3) space/extension of incoming connection can be received to define node wherein by defining the distribution of I/O points, and define node energy wherein Space with outflow connection.

For incoming extension, it is possible to specify receive the distribution of end.The distribution for receiving end can be similar to neuron situation In dendron (dendrite) territory, and/or synaptic knob distribution.For outflow extension, it is possible to specify projection territory.Projection territory Aixs cylinder (axon) end can be analogous to.

Distribution for receiving end and projection territory, HLND can support pre-defined function (for example, bounded gaussian sum/ Or be uniformly distributed).In general, any density function can be used.

Exemplary HLND is invoked in list below 2 and shown：

Example1_exc_neurons_axon=(' example1_exc_neurons ', pdf1)

Example2_exc_neurons_dendrite=(' example2_exc_neurons ', pdf2)

List 2.

In another method, connection be instantiated and without using form/extension.

Connecting node

HLND connection sentences may include to be configured to instantiate (the given connection class from a set of node to another set of node Type) instruction of connection.In some implementations, it may be required that some or all of to realize these nodes extremely in following information Node is connected：(1) " coming from subset ", (2) " going to subset ", and/or (3) " connection type "." coming from subset " may include by only Identify to one property source node/extension (for example, connection by from node/extension) the selected Node subsets of mark." go Toward subset " it may include by the node/extension (for example, connection by the node/extension ended in) of uniquely recognition purpose ground Mark selected Node subsets.Connection type may include to be used to incite somebody to action<From subset>Node/extension is connected to<Go to subset> The connection type of node/extension.

In some implementations, HLND connections sentence can be with from all available<From mark>Node is to all available <Go to mark>The given connection type of node instantiates to instruct to connect.According to some realizations, Connecting quantity can be used for filter Except connection.That is, filter constraint can be applied to some or all of possible<From mark>Extremely<Go to mark>Connection.Therefore, It is possible<From mark>Extremely<Go to mark>The subset of connection can be instantiated, this can allow from<From mark>Extremely<Go to Mark>Any connection mapping instantiation.In some implementations, connection can be expressed as function call.In some realizations In, connection can be used form to express.In some implementations, connection filter can be configured to all to all connections of generation, It is wherein all<From mark>It is connected to all<Go to mark>.

Incidentally, can be in various implementations using two kinds of annotation forms<From mark>With<From subset>, because this Two kinds of annotations can make HLND generate connection for subset.For example, annotation<' come from mark 2 ' from the AND ' of mark 1 '>It can retouch State with mark ' from mark 1 ' and ' set (for example, set 1) of the node from both marks 2 '.Correspondingly, it can replace Ground uses annotation：<' set 1 '>And produce identical result.

By non-limiting example, following connection sentence can be used：

Exc2exc=(pre='example1_exc_neurons_axon',

Post='example2_exc_neurons_dendrite', 100connections/cell,

SynType='GLU', _ other_parameters_).

In some implementations, HLND connections sentence can be configured to realize that the connection of parametrization is set up, so that each parameter can Connection type is passed to set the link variable in connection example.In some implementations, Connecting quantity can be used for setting prominent Touch the weight of node connection.In some implementations, nodal information (for example, position of " coming from node " and " going to node ") can quilt For setting up connection weight based on the distance between node.

, can be using one or more of following connection sentence by non-limiting example：(1) will be each<From section Point>Node is connected to N number of<Go to node>Node；(2) will be N number of<From node>Node is connected to each<Go to node>Section Point；And/or (3) are N number of from stochastical sampling in connection is possible to.

In some implementations, node may include position mark and/or can have zero acquiescence extension.Such node can connect It is connected to coexistence node.Crossover is may require come connecting node with free token, so that crossover node can be connected.

Referring now to Fig. 3 A, the exemplary realization of HLND frame joints interconnection is shown and is described in detail.Fig. 3 A network 300 may include A node clusters 302 and B node group 304.For clarity, one-dimensional configuration can be used to configure for network 300, so as to have (that is, Fig. 3 A interior joints index i=1 of X-coordinate 306 of matching:7) each node is allowed to connect via connection 308.Specifically Ground, as i=j, (node 312 for example, respectively in Fig. 3 A, 314) can allow from node cluster 302 Node a_i is connected to the respective nodes b_j of node cluster 304.Some possible connections are illustrated by solid line 308.

In some implementations, such as explained in Fig. 3 B-3C, point spread can be added separately to each node cluster Node, to realize complicated connection and to realize more flexible connection mapping.Point spread can be particularly useful for source node (example Such as, the node A of the group 322 in Fig. 3 B) space coordinate be mapped to receiving node (for example, node 324 in Fig. 3 B).These expand Exhibition can be used for the probability density function for defining potential connection, in the exemplary realization such as explained in fig. 3 c like that.

Fig. 3 B illustrate the exemplary realization that the node including point spread connects configuration to node.Fig. 3 B network 320 It may include A nodes a_1:A_7 group 322 and single B node 324.Single dimension can be used to configure for Fig. 3 B network 320, from And each node of the X-coordinate 306 with matching can form connection.Fig. 3 B B node 324 may include with the equal of dimension 332 Even dendron extension 330.The individual nodes (such as node 322_1) of A node clusters 322 may include the uniform aixs cylinder with dimension 328 Extension 326.Aixs cylinder dimension 328 is smaller than dendron dimension 332.Term " uniform expansion ", which can be used for description, to be used to realize that crossover expands Open up the non-uniform probability distribution of the connection of (aixs cylinder or dendron in this example).That is, according to some realize, node connection can by with Impartial possibility provides it and extends crossover.For One-Dimensional Extended (for example, extension 330 in Fig. 3 B), this may correspond to along expansion The uniform connection probability for extending to scope 332 of exhibition 330.For Multi-Dimensional Extension, uniform expansion may correspond in all dimensions Uniform connection probability.

When 332 crossover of Spatial Dimension of aixs cylinder 326_i and dendron 330, the network configuration explained in Fig. 3 B can allow A Connection between node a_i and the b_1 node 324 of group 322.As explained in Fig. 3 B, node a_3, a_4, a_5 can be via these It is a little to connect to be connected to node b_1, as the solid arrow 308 in Fig. 3 B is described.Inactive (example between other A nodes Such as, it is not allowed to) connection 318 described by dotted arrow in figure 3b.Mark<' aixs cylinder '>With<' dendron '>It may refer to node Another ' view '.

As explained in Fig. 3 B, following extension can be constructed：(i) the Homogeneous Circular extension of the B node 330 into Fig. 3 B (being expressed as dendron)；The Homogeneous Circular extension (being expressed as aixs cylinder) of all A nodes of (ii) into Fig. 3 B；And/or (iii) will<A Aixs cylinder>It is connected to<B-tree is dashed forward>.A is possible and can be instantiated to B node connection, and both are extended because sending and receiving Can be uniform.In the example present, connection sentence is searching extension crossover --- i.e., node<A>'s<Aixs cylinder>Extension Whether with node<B>'s<Dendron>Extend crossover.

In HLND frameworks, each node may include different views, for example, aixs cylinder or dendron.Vertex ticks ' aixs cylinder ' or ' dendron ' can be used to refer to same node in HLND.Aixs cylinder/dendron can have different spatial properties.

Fig. 3 C illustrate the exemplary realization that the node extended including non-uniform knots connects configuration to node.Retouched in Fig. 3 C The network 338 painted may include group 322 and the B node 344 of A nodes.Individual nodes in these A nodes may include uniform expansion 326.B node 344 may include non-homogeneous extension 340.The non-homogeneous extension of term can be used at least one dimension of description across extension Point spread (aixs cylinder or dendron) of the degree with non-homogeneous connection probability distribution.For One-Dimensional Extended (for example, the extension in Fig. 3 C 340), this corresponds to the non-homogeneous extension connection probability distribution that scope 342 is extended to along extension.In some implementations, extension connection Property parameter P may include to connect likelihood, it can be characterized by the probability function for the function for extending to as extension scope 342.

Gaussian Profile and/or the shape of other distributions can be used in the connective section view (referring to Fig. 3 C) of non-homogeneous extension 340 To configure.Extension 340 can be centered on node 344.Extension 340 can be by particular variance σ²Characterized with radius 348.

Explained in fig. 3 c via the node connection of non-homogeneous extension.When aixs cylinder 326_i and non-homogeneous dendron 340 space During 342 crossover of dimension, the network configuration explained in Fig. 3 C can allow the company between node a_i and the b node 344 of A groups 322 Connect.In network 338 possibility connection include a_3, a_4 ..., a_10 aixs cylinders to b dendrons because a_1 aixs cylinders, a_2 aixs cylinders and a_ 11 aixs cylinders not with the crossover of 340 spatial spread of dendron 342.

Non-homogeneous extension (for example, extension 340 in Fig. 3 C) can make connection selection deviation and the highest of non-homogeneous dendron The aixs cylinder of probability region crossover.Although extension dimension crossover can be used for identifying all possible connection, to possible connection Sampling can follow the connective section view (probability) of extension.When selecting the subset of possible connection, it may be necessary to which considering should (description connection likelihood) connective section view (shape) of extension.By non-limiting example, when will be single any<A aixs cylinders> 326 are connected to<B dendrons>(referring to Fig. 3 C) when 340, most possible result is probably the company between node a_6 and node 344 Connect, and the connection between not a node a_1, a_10 and node 344.

In some implementations, can be with for carrying out the HLND exemplary sequences of operations of connecting node colony using non-homogeneous extension It is：(1) it is by the Gaussian spread with radii fixus r1 centered on node added to all B nodes and by these extending markings <Dendron>；(2) there will be radii fixus r₂Uniform expansion added to all A nodes and by these extending markings be<Aixs cylinder>；With And (3) will be N number of random<A aixs cylinders>It is connected to<B-tree is dashed forward>.Possible connection can be<A>Node<Aixs cylinder>Extension with<B> Node<Dendron>Extend crossover part.It is connective that the connection being instantiated may correspond to highest.In some implementations, highest connects The general character can be determined based on the product of gaussian sum uniform function.

I/O for network object

In some realizations of HLND frameworks, network object may include one or more members and input/output (I/O) Interface.The I/O interfaces can be specified how other elements docking (being connected for example, setting up) with network.In some implementations, network The member of object may include node.In some implementations, the member of network object may include node and connection.I/O interfaces can be with How definition is can be from the outside access object member of the scope of network object.During defining, the individual member of network object (and Its value) can be declared as it is public or privately owned.Privately owned member may be invisible for the external network element outside the network object (that is, not directly access).Private object member can access via for the I/O interfaces of the member definition.That is, network object Privately owned member may be invisible outside the scope of the network object.In this case, it may be required that I/O interfaces are realized Connection.

In some implementations, network object can be defined as what is ' opened '.The member for being defined as open network object can To be public and outside the scope of the network object.This can mitigate the requirement for announcing I/O interfaces.

The public member of network object can be visible and/or addressable for input and/output company by external elements Connect.In some implementations, the member of network object, which can give tacit consent to, scope.That is, some or all of variables in network object Scope can be limited in the network object.The member of multiple examples of consolidated network object type uses phase in these members Interference will not be produced during isolabeling.

In some implementations, network object can be defined as ' grand '.Being defined as grand network object can not be by as having The object of scope is treated.Such a macrodefinition can allow some or all of variables defined in grand object can by external elements Access and/or visible.

By non-limiting example, node _ a can be network object NO1 public member, and (it is not NO1 to node _ b Member) node _ a is can be directly connected to, and/or receive the connection from node _ b.Can be by using mark<NO2>With<Section Point _ a>, and/or have scope annotation to access member node _ a in NO1 with NO1. nodes _ a.According to some realizations, net is used as Node _ c of network object NO2 privately owned member may from outside invisible and/or inaccessible, unless for NO2 node _ c into Member defines I/O interfaces.External node (it is not NO2 member) is not directly connected to node _ a member and/or directly connect Receive the connection from node _ a member.In other words, mark and/or other publicly available information can be used to access network object Public member.

The exemplary reality for the HLND network objects for explaining public many chamber neurons (MCN) has shown and described with reference to Fig. 4 It is existing.MCN neurons 400 may include one or more common nodes, and the one or more common node may include dendron chamber 404 With cell space chamber 402.The public scope outside (example from its definition that may refer to network object (for example, chamber 402,404) of term Such as, outside MCN 400) visible member.Individual chamber 404 can be assigned two marks, and the two marks may include DENDRITE (dendron), COMP (chamber), and/or other marks.Chamber 402 can be assigned three marks, and these three marks can Including DENDRITE, COMP, SOMA (cell space), and/or other marks.Chamber 402 and individual chamber 404 can be connected via knot 406 connect.

By non-limiting example with explain according to some realization public network elements feature, it may be considered that it is public Two examples of MCN neurons 400.One example can be marked as ' neuron a ', and another example can be marked as ' nerve First b '.Annotation<Neuron _ a AND cell spaces>Set may refer to the MCN member that there is cell space to mark in neuron _ a examples 402.<Neuron _ b AND cell spaces>Set may refer to the MCN member 402 that there is cell space to mark in neuron _ b examples.Because MCN<Neuron _ a>Some or all of members 402,404 can be public, so they for external entity (for example, MCN<Neuron _ b>) can be visible, this, which may be such that, to realize<Neuron _ a AND cell spaces>Extremely<Neuron _ b AND Dendron AND chambers>Be directly connected to.

In some implementations, from<Neuron _ a AND cell spaces>Extremely<Neuron _ b AND cell spaces>Connection can be used it is given Connection type instantiate.

As will be understood as those skilled in the art, above annotation is exemplary and can use various other annotations To be identified, selected using the mark of node member, and/or accessed node member.

Fig. 5 illustrates public neuron A and B according to one or more realizations (in Figure 5 respectively by the He of designator 500 520 indicate) between connection instantiation.Node A individual member 504 and node B individual member b 524 can be public 's.Individual ' a ' member 504 can be connected to individual ' b ' member 524.Illustrate an example in Fig. 5, wherein outer process 508 can be with The crossover of multiple inner processes 526 of node 524, so as to the company of foundation between single a nodes 504 and one or more b nodes 524 Connect 508.

Given in Fig. 6 with explained in Fig. 5 realize corresponding exemplary pseudo code.Sentence 600 and 620 in Fig. 6 can It is configured to generate Fig. 5 node instance 504 and 524 respectively.Sentence 606 and 626 in Fig. 6 can be configured to definition figure respectively 5 outer process 506 and/or inner process 526.Last sentence 610 can be configured to definition connection 508.

Fig. 7 illustrates the exemplary realization of the network object including privately owned many chamber neurons.Privately owned MCN neurons 700 It may include one or more privately owned chambers 704 and/or privately owned chamber 702.Term ' privately owned ' may refer in corresponding network object Outside (for example, MCN 700 outside) sightless network memberses (for example, chamber 702 and 704) of the range of definition.Individual room Room 704 can be assigned two marks, and the two marks may include DENDRITE, COMP, and/or other marks.Chamber 702 can quilt Three marks are assigned, these three marks may include DENDRITE, COMP, SOMA (cell space), and/or other marks.

Explained by non-limiting, it may be considered that two examples of the privately owned types of MCN 700.One example can be marked as ' neuron _ a ', and another example can be marked as ' neuron _ b '.Because the types of MCN 700 are defined as privately owned, MCN 700 members (for example,<Neuron _ a AND dendron ＆ chambers>、<Neuron _ a AND cell spaces>、<Neuron _ b＆ dendron ＆ chambers>、 <Neuron _ b＆ cell spaces>、<Cell space>Set) can be sightless from MCN 700 outside.According to some realizations, for Fig. 7 The node Configuration Type of middle explanation, directly will may not allow<Neuron _ a AND cell spaces>It is connected to<Neuron _ b AND trees Prominent AND chambers>.In order to realize neuron exterior connectivity, the definition of MCN 700 may include input (IN) port 714 and 716, with And output (OUT) 718 port, it can be used for specifying the I/O interfaces for the MCN node types.MCN 700 input interface It may include to the direct internal connection of MCN 700 privately owned member.Direct internal connection to MCN 700 privately owned member can be wrapped Include the connection 726,728 and 730 from input interface IN1 714 to the privately owned member 704 with mark " dendron " and " chamber ".Extremely The direct internal connection of MCN 700 privately owned member may include from input interface IN2 716 to the privately owned of mark " cell space " The connection 720 of member 702.Privately owned member 702 can be connected to output (OUT) interface 718 by link 722, and this can allow node 702 It is used for outflow connection and/or for incoming connection.Neuron _ a.OUT can be configured to link/be connected to neuron _ b.IN1.Neuron _ b.OUT can be configured to link/is connected to neuron _ a.IN2.

Fig. 8 illustrates the example between private network object A and B (being indicated respectively by the designator 800 and 820 in Fig. 8 A) Property connection instantiation.Individual member in node A member 804 and/or the individual member in node B member 824 can be Privately owned, and therefore can not may be accessed by external instances.That is, realize that ' a ' node member 804 can be not according to each It can be accessed by ' b ' node member 824, vice versa.In order that node instance 800 can be generated to the external connection of node 820, The I/O interfaces for privately owned node member can be required.In some implementations, I/O interfaces may include input/output end port (example Such as, I/O ports 714,716 and 718, are described above with respect to Fig. 7).Privately owned node instance (for example, example 800) may include greatly Measure other privately owned members of (for example, millions of).Privately owned node instance can provide for those members will be used for outflow connection Outgoing interface 812.

Conversely, according to some realizations, although it is privately owned that can keep the individual member in the member 824 of node instance 820, but Input interface 822 can be used for specifying how node member 824 is connected to input port 822.Although being explained in Fig. 8 A realization Single port, but this is not intended as composition and limits because can in some implementations using it is multiple uniquely marked it is defeated Enter/output port.

Because the output interface 812 of node instance 800 and the input interface 822 of node instance 820 can be externally exposed net Network element and/or visible to outside network element, it is possible to by using input and output interface 812 and 822 with similar to Above with respect to Fig. 5, Fig. 7 describe connection set up mode set up between the node instance 800 and 820 of network object link/ Connection 810.

The connection that Fig. 8 B are given with being explained in Fig. 8 A, which is set up, realizes corresponding exemplary pseudo code.Sentence in Fig. 8 B 830 can create the node instance 804 in Fig. 8 A and can expose output port 812.Sentence 850 in Fig. 8 B can be created Node instance 824 in Fig. 8 A.Sentence 856 and 852 in Fig. 8 B can define the inner process of Fig. 8 A ' b ' node 824 respectively 826 and/or Fig. 8 A input interface 822.Sentence 858 can define the connection between the input interface and ' b ' member.

In some implementations, type A 800 and B 820 example inst_a can be respectively created in the 3rd network object 860 And inst_b, and/or the a_ output ports of node instance 800 and the input port of node instance 820 can be used by inst_a Example is connected to inst_b examples.The exemplary HLND definition steps shown in Fig. 8 B may include that (1) creates A example (inst_ A), (2) create B example (inst_b), and/or by inst_a connections and/or are linked to inst_b.

Because node instance A and B member are probably privately owned, object C is possibly can not be direct by example A member It is connected to example B member.Replace, object C can use exposed port inst_a.a_out to inst_b.in. In connection, statement 868 can use equal sign annotation and be assigned to (for example, being same as) inst_a.a_out to indicate inst_b.in. Some realize in, HLND compilers can use define 868 and privately owned node B member definition by by inst_b.in with section Point example A 800 correspondence (all) member links (that is, the connection set up indirectly between (A's) inst_a and (B's) inst_b) To parse the connection between virtual b.in ports and node B 820 actual member.As explained in Fig. 8 A, node type 800 and 820 may specify to carry out node to the be may require projection extension of node connection and/or synapse type (SynType)/connection type.

Mark

According to some realizations, the individual element of volume (that is, node, extension, connection, I/O ports) of network can be assigned at least one To facilitate, HLND is operated individual uniqueness mark and ambiguousness is eliminated.Each mark can be used for identifying and/or quote from corresponding network Element (for example, Node subsets in designated area of the network).

In some implementations, mark may be used to form the dynamic clustering of node, so that the node cluster of these dynamic creations It can be connected to each other.That is, node group mark can be used for identifying the new connection in Node subsets and/or establishment network, such as following It is described in detail with reference to Fig. 9.These additional markings can not be the new example for creating network element, but can add mark To existing example, so that additional marking be used to identify labeled example.

Fig. 9 illustrates using additional marking to identify the exemplary realization of labeled example.Network node colony 900 can wrap One or more nodes 902 (being labeled as ' my node '), one or more nodes 904 are included (labeled as ' my node ' and ' son Collection '), and/or other nodes.The signable mark of dark triangle in Fig. 9 node colony 900 is my node ' node 902, and black and white triangle may correspond to the Node subsets 904 labeled as ' my node ' and ' subset '.

Use mark ' my node ', it is possible to select node set 910.Node set 910 may include node 902 and/ Or all individual nodes in 904 (see, for example, Fig. 9).Node set 920 can represent to be labeled as<' my node ' NOT ' Collection '>Node.Node set 920 may include all individual nodes in node 902.Node set 930 can represent to be labeled as The node of ' subset '.Node set 930 may include all individual nodes in node 904 (see, e.g. Fig. 9).

In some implementations, HLND frameworks can use two kinds of mark, its may include string mark, numeric indicia, And/or other marks.In some implementations, each node may include the mark being arbitrarily defined by the user.Numeric indicia may include Numeric identifier (ID) mark, free token, and/or other marks.

When instantiated nodes, the node being instantiated can have string mark (node type) and uniqueness numerical value mark Remember (uniqueness value identifiers).In some implementations, position mark can be assigned during instantiation process.

Computing to mark

As shown in Figure 9, each mark can be used for the subset for identifying network., can be to mark in order to realize this feature Note uses one or more Boolean calculations.In some implementations, mathematical logic computing can be combined with numeric indicia.<…>Annotation can To identify the subset of network, wherein by angle brackets<>The string of encapsulation can define the operation for being configured to identify and/or selecting subset. Explained by non-limiting,<' my mark '>All individual nodes with mark ' my mark ' can be selected from network；< ' my marks 2 ' of my AND ' of mark 1 '>It can be selected from network with mark ' my mark 1 ' and ' my string of mark 2 ' Both all individual nodes of mark；<' my marks 2 ' of my OR ' of mark 1 '>Can be selected from network with mark ' I Mark 1 ' or ' all individual nodes of my string mark of mark 2 '；<' my marks 2 ' of my NOT ' of mark 1 '>Can be from network Selection have string mark ' my mark 1 ' but without string mark ' all individual nodes of my mark 2 '；And<' my mark 1 ' AND MyMathFunction (free token)<Numerical value 1>Can be selected from network have string mark ' my mark 1 ' and Numerical value 1 is less than by MyMathFunction (my mathematical function, when being employed the space coordinate of the node) outputs provided All individual nodes.Note, according to each realization, this example assumes Existential Space mark, and this is not compulsory.

Mark is inherited

In some implementations, HLND frameworks may include hierarchy type mark succession.In some implementations, it is real in network object The individual member of exampleization can inherit the string mark of its father.For example, network object has mark ' father's mark 1 ' and ' father's mark All individual members of note 2 ' in addition to the mark different because of member assigned during member instantiates except for example may also include mark ' father's mark 1 ' and ' father's mark 2 '.

In some implementations, passenger compartment's flag data may refer to the local coordinate of the member (with reference to being determined by network object The space of justice).In some implementations, world coordinates (with reference to the space of whole network) can be from network object and/or member It is inferred in nested structure.

Figure 10 illustrates the exemplary realization of free token succession.Network object B (not shown) can instantiate Type C (for example, node _ c at position (1,1) place, 1002) it be denoted as the single instance of node in Fig. 10.Network object A (not shown) can with two examples of instantiated nodes type B (for example, position (1,1) place node _ b_1 and position (1, 2) node _ b_2 at place, it is denoted as 1004 and 1006) respectively in Fig. 10.Node _ c coordinate can refer to node _ b_1's Scope.Node _ b_2 coordinate can be set as (1,1).The coordinate of node _ b_1 and node _ b_2 reference node _ a scope can It is respectively set to (1,1) and (1,2).The coordinate of the scope with reference to node _ a of node _ c in node _ b_1 can be confirmed as (1,1)+(1,1)=(2,2).The coordinate of the scope with reference to node _ a of node _ c in node _ b_2 can be confirmed as (1,2)+ (1,1)=(2,3).Note, annotation node _ c, node _ b_1, node _ b_2 and node _ a can be used for identity type C, B respectively With the A object being instantiated.

Realized according to one or more, HLND scalar natures and/or characteristic can be summarized as follows：Type may include String mark and numeric indicia；Numeric indicia may include value identifiers；Boolean calculation can be used for mark；Logarithm value mark can be permitted Perhaps mathematical function；Optional free token and string can be assigned to mark；Individual nodes example may include uniqueness value identifiers mark Note；It can be hierarchy type that string mark, which is inherited,；Free token may refer to local coordinate；Global mark coordinate can be from the embedding of node It is inferred in nested structure；And/or other properties and/or characteristic.

Mark is realized

In some implementations, the HLND frameworks of mark are realized and can be configured to require following functions：(i) to mark-number According to maker and the interface of data processor；The realization of (ii) nested object is so that can be from any amount of existing network Object Creation complex network object.

In some implementations, flag data processor can use database (for example, MySQL) to realize. The example of network object can use any string mark to generate.In some implementations, network object can use position mark And string marks to generate.Flag data can be placed in database.Can generate complementary network data (for example, connection example, Knot, cynapse etc.).The instantiation of connection can depend on position mark and/or Query Result.New data can be stored in number According in storehouse.

Mark realizes that configuration can make it possible to network software application being divided into two parts, and it may include that data are generated Block, DSB data store block, and/or other parts.Data generation block (for example, being realized with C++) can be configured to based on its own ' intelligence ' and/or generate data by being interacted with database (for example, MySQL).In some implementations, Data Generator function Property can be embedded in database server.The server side code that Data Generator can use by triggering to activate is come real It is existing.Such triggering may include to be stored in insertion and call connected/triggering code on database server.

In some implementations, instantiation END cynapses/knot may require the information of one or more etc below such as：It is prominent Touch front unit 1 class and ID, the class of postsynaptic unit 2 and id, the locus of cynapse front unit 1 and postsynaptic unit 2 and Cynapse front unit 1.out and postsynaptic unit 2.in space projection, and/or other information.

Cynapse/solid example can be generated.In some implementations, additional external parameter can be used for the reality of END cynapses/knot Exampleization.The example of external parameter may include synapse weight, synaptic delay, and/or other external parameters.During node is connected to node Synapse weight and/or the use of delay and feature be to submit June 2 for 2011 it is jointly owned entitled " APPARATUS AND METHODS FOR TEMPORALLY PROXIMATE OBJECT RECOGNITION (are used for the time The apparatus and method of the identification of close object) " U.S. Patent application No.13/152,105 and/or on June 2nd, 2011 Jointly owned entitled " the APPARATUS AND METHODS FOR PULSE-CODE INVARIANT OBJECT submitted Enter in RECOGNITION (being used for the apparatus and method of the constant Object identifying of pulse code) " U.S. Patent application No.13/215,575 One step is described in detail, and above-mentioned each full content is incorporated by reference thereto.

In order to connect the heterogeneous networks object using different spaces coordinate, the coordinate for each network object can be announced System.That is, coordinate system configuration can use to the individual nodes in the range of some.

In HLND frameworks, the connection between network object can be set up by one or more modes.In some implementations, The connection can be distributed the crossover between synaptic knob distribution based on axon ends to set up.In some implementations, overall connection is reflected Penetrating can use the joint probability distribution function (PDF) of axon ends distribution and synaptic knob distribution to obtain.Combined PD F can by with In the connection (cynapse) needed for foundation.In some implementations, HLND frameworks can be configured to the individual being distributed in potential tie point Tie point.Tie point can be limited by one or more specified conditions, such as space coordinate and/or other conditions.HLND connections are calculated Method can be configured to select all (or subsets) of these tie points.HLND join algorithm can be configured to instantiate corresponding company Connect.In some implementations, HLND can be configured to generate the connection collection being arbitrarily defined by the user.According to some realizations, HLND It can be configured to all to all connections of generation.

Class SQL forms

In some implementations, HLND can be realized using SQL completely.According to some realizations, such a SQL, which is realized, to be made Carried out with MySQL database and the function/code stored.HLND sentences can be constructed according to English language grammer.

Mark example

As described above, (no matter it is node, unit, cynapse or knot or is only the individual networks element defined in HLND Empty placeholder) it may include mark.This characteristic of HLND networks description can allow labeled element especially as group's quilt Addressing and manipulation.In some implementations, space coordinate can use the mark of (x, y, z) form to realize.

By non-limiting example, NE may include one or more marks, it include unit ID value identifiers, ' QIF ', ' cell space ', ' pyramid ', ' layer 2/3 ', ' V1 ', space coordinate mark (0.1,0.3,0.5), and/or other marks.It is prominent Such as following mark can be had by touching：UD, indicate respectively before the neuron of presynaptic and postsynaptic node ID, after neuron and/ Or other marks, ' top ', ' exc ', ' glu ', and space coordinate mark (0.1,0.3,0.4).

Tagged (labeled) operators and label filter

In some implementations, quick access elements group can be allowed by mark being stored in database.Mark is transported The individual data items library inquiry sentence of calculation may act as selecting the mark filtering of (being matched with query term) element-specific from database Device (or search statement).For example, specifying in queries<'V1'>Can cause selection include in its any mark ' V1' it is individual Element of volume, such as whole V1 subsets.Specify (<' V1'AND' pyramids ' NOT' layers of 2/3' of AND) can obtain in V1 not positioned at net Individual pyramid cell in network layers 2 and 3.

In some implementations, the output of tag query can be assigned the mark of their own as follows：

<tag filter>TAGGED new tag

List 3

Some realize the element that addressing can be allowed to meet label filter (tag filter), without replicating and pasting Filter sentence.

Example 3.

Following sentence：

exc OR inh TAGGED all

Mark ' all ' (all) can be easy to added to all ' exc ' and ' inh ' neuron quote.

Example 4.

Following sentence：

(exc AND id<400)OR(inh AND id<100)TAGGED first_half

Network usually can be cut to two halves by the way that additional markers are assigned into the first half (first_half) member.

OF operators and subset

In some implementations, expression formula

n OF<tag filter>

List 4

The list for n random element for meeting label filter (tag filter) condition can be returned.If labeled Filter returns to less than n element, then some elements can be replicated, so that the total quantity of the element returned by the expression formula of list 5 Equal to n.OF operators can not assign new mark.OF operators can select subset of elements.Marked to be assigned to the element of subset Note, can use TAGGED operators.Expression formula

100OF cones TAGGED S_cones

List 5.

100 elements can be selected from the node colony of the cone (cone) and individual chosen elements can be labeled as S_cones.Similarly, expression formula

300OF(cones AND NOT S_cones)TAGGED M_cones

cones AND NOT M_cones AND NOT S_cones TAGGED L_cones

List 6.

300 elements can be selected from the node colony of (not in the subset S_cones) cone, can be by individual choosing Rubidium marking is determined for M_cones, can be from the section of (neither in S_cones subsets nor in the M_cones subsets) cone Remaining individual element is selected in point colony, it is possible to which each chosen elements are labeled as L_cones.

Example 6.

According to some realizations, include the network of 800 excitability (exc) neurons and 200 inhibition (inh) neurons Two equivalent sub-networks, i.e. sub-network 1 (subnetwork 1) and sub-network 2 (subnetwork2) can be split as, its is each It is as follows from 400 exc and 100 inh neuron selected at random is included：

400 OF exc TAGGED subnetwork 1

100 OF inh TAGGED subnetwork 1

400 OF(exc AND NOT subnetwork 1)TAGGED subnetwork2

100 OF(inh AND NOT subnetwork 1)TAGGED subnetwork2

List 7.

The realization of list 7 is contrasted with following sentence：

500 OF(exc OR inh)TAGGED subnetwork1

500 OF(exc OR inh)TAGGED subnetwork 2。

List 8.

The sentence of list 8 does not ensure that each of sub-network 1 and sub-network 2 accurately excite nerve including 400 Member and 100 inhibitory neurons.

The instantiation of PUT operators and unit

The PUT operators can be used for instantiating as follows and token network unit：

PUT n OF unit_class

List 9.

PUT operators can be create ' the n example of unit _ class (unit_class) ' type and be marked as (id, Unit_class instruction).Additional marking then can be added into these units by using TAGGED operators.PUT operators can With the relative configurations device function of call unit _ class to instantiate individual cell.In list 10, OF keywords can be used for so that N copy of the unit _ class is generated by call unit _ class constructor n times.

Example 8：

According to some realizations, sentence

PUT 800 OF exc

List 10.

800 units of exc classes can be created, wherein individual instances are marked as (id, exc).

Following sentence

PUT 800 OF exc TAGGED exc_neurons

List 11.

800 units of exc classes are created, wherein individual instances are marked as (i) (id, exc)；And/or (ii) additional mark Exc_neurons is remembered, so that individual instances include two marks.

In some implementations, PUT operators can be used to carry out creating unit example by using filter parameter as follows：

PUT<tag filter>OF unit_class

List 12.

The instruction of list 12 can be configured to create unit_class with by<tag filter>(label filter) is looked into Ask the example of the corresponding quantity of the selected number of elements of field.The individual unit being instantiated can use and unit in list 12 The respective element selected in list by the inquiry corresponding mark is marked.When constructed fuction unit_class is called, It can be able to access that the mark (for example, ID, coordinate etc.) of the element of the example, so that there is constructor sufficient information to use In unit construction.

Example 9：

Sentence

800 ON circle (l) TAGGED my_points//referring to definition below to ON

PUT my_points OF exc

List 13.

It can be configured to instantiation and distribute 800 units of exc classes on unit circle (circle).Can be by using conjunction Reach identical result into the OF exc ON circle (1) of sentence PUT 800.

CONNECT operators and connection unit

Connection (CONNECT) operator can be used for Synaptic junction as follows：

CONNECT pre_tag TO post_tag WITH synapse_class

List 14.

Wherein parameter synapse_class (cynapse _ class) signs are defined for the class of Synaptic junction, and preceding mark (pre_ Tag) indicate with rear mark (post tag) and specify the filter of the cynapse front unit connected by cynapse and postsynaptic unit to cover Code.In some implementations, multiple presynaptic and/or postsynaptic units can be selected by filter mask, thereby cause meeting Multiple Synaptic junctions are generated between the unit of filter mask.

In some implementations, synapse_class can in CONNECT sentences by junction_class (knot _ class) Lai Instead of so as to generate synaptic knob.Constructed fuction synapse_class can be able to access that presynaptic and/or postsynaptic member The individual mark of element.Constructed fuction synapse_class may decide that delay and/or have related parameter needed for other.

Example 9.

Some realizations can provide following sentence：

CONNECT N OF pre_tag TO post_tag WITH synapse_class

CONNECT pre_tag TO N OF post_tag WITH synapse_class

List 15.

The first sentence in list 15 can be configured to generate connection matrix, so that individual post_tag (rear mark) unit It is connected to N number of cynapse front unit.It is single from individual pre_tag (preceding mark) that the second sentence in list 15 can be configured to generation N number of outflow cynapse of the member to n post_tag unit.

Sentence in list 15 can use randomly selected subset.This can be by all units for having added preceding mark Random walk and randomly choose the member of subset in list and usually realize.

Example 10.

Some realizations can provide following sentence：

n OF(CONNECT pre_tag TO post_tag WITH synapse_class)

CONNECT pre_tag TO NEAREST post_tag WITH synapse_class

CONNECT NEAREST pre_tag TO post_tag WITH synapse_class

List 16.

The first sentence in list 16 can be configured to the random subset that instantiation includes the full-mesh matrix of anterior-posterior cynapse Cynapse.Term full-mesh matrix can be used for describing the network that all cynapse front units are connected to all postsynaptic units Configuration.Different from the example shown in list 15, the first sentence in list 16 does not ensure all cynapse front units or all prominent Unit includes the cynapse of identical quantity after touch.

In list 16 second and/or the 3rd sentence can be configured to generation based on cynapse front unit and postsynaptic unit The Synaptic junction of coordinate.Second sentence may include to be configured to be connected to each cynapse front unit meet mark mask near Postsynaptic unit ring.3rd sentence with the every post_tag of searching loop and can find hithermost (NEAREST) pre_ tag。

In some implementations, parameter NEAREST 1OF can be used for the parameter NEAREST in the sentence instead of list 16.

In some implementations, individual cynapse front unit can use following sentence to be connected to n hithermost postsynaptics Unit：

CONNECT exc TO NEAREST n OF exc WITH glu

List 17.

It can create a hithermost excitability from individual excitability neuron (that is, labeled as the unit of ' exc ') to n ' glu ' type cynapse of neuron, this n hithermost excitability neurons include the individual excitability neuron itself (i.e., Cause one from cynapse).

The OF operators of popularization

In some implementations, the OF selection opertors of popularizing form can be configured as：

<tag filter 1>OF<tag filter 2>

List 18.

The OF operators of popularization can perform it is following one or more：(i) create and meet mark wave filter 1 (tag filter 1) list 1 (list_1) of all n elements of condition；(ii) create and meet mark wave filter 2 (tag filter 2) condition All m elements list 2 (list_2)；(iii) created by (randomly) selecting the subset of n element from list 2 List 3 (list_3) is built, if n>M, then have n-m element to be repeated in list 3；(iv) return and merge list, wherein coming There is the additional marking of the coupling element in list 3 from each element of list 1；And/or other actions.If list 1 Both include coordinate with list 3 to mark, then the individual element of volume in merging list may include the coordinate summation for corresponding element Coordinate is marked, so as to maintain the single coordinate (x, y, z) of every element to mark.

Example 11.

In some implementations, available ' cone (cones) ' is marked unit collection.Random coordinates collection can use ' retina (retina) ' mark.Following formula can be used to be assigned to the cone for random retina coordinate：

cones OF retina

When the quantity of the cone is more than the quantity of coordinate, then multiple cones can be assigned identical coordinate.

ON operators and coordinate are assigned

ON operators can be used for the sampling for returning to n points from the probability density function defined by parameter pdf as follows：

n ON pdf,

Example 12

Some realizations can provide following sentence：

1000 ON segment(0,1)TAGGED rnd

1000 ON circle(1)TAGGED cones

List 19.

First sentence of list 19 can be configured to list of the generation labeled as the element of ' rnd '.These are labeled as ' rnd ' Element can be evenly distributed in the space segment (segment) defined by coordinate x=[0 1].That is, element can have mark (rnd, x), wherein x values are evenly distributed in scope [0 1].The second sentence in list 19 can be configured to generation and be labeled as The list of ' cones ' (cone) and 1000 elements being evenly distributed on unit circle.

In some implementations, ON operators can use the individual point returned by function F：

ALL ON F

PER operators and mark are combined

Operator PER can be used for being iterated through the mark list (tag_list) specified by such as label filter. For the individual element of volume of the list, operator PER can be with call statement (statement), so as to transmit list element to it All marks.The form of PER operators can be：

statement PER tag_list

PER operators can return to the table for including the data for describing generated network element.In some implementations, PER is calculated Son can be used for the multiple cynapses for creating every neuron.In some implementations, PER operators can be used for creating every position (location) multiple neurons.In some implementations, PER operators can be used for the multiple cortexes for creating every cortical surface Row.

Example 13

Some, which are realized, provides following sentence：

1000 ON segment PER neuron

1000 OF locations PER neuron

List 20.

SPNET

In SPNET some realizations are gone for, unit class exc and inh and cynapse class glu and gaba can be Defined in SPNET definition.

PUT 800 OF exc

PUT 200 OF inh

CONNECT exc TO 100 OF exc OR inh WITH glu

CONNECT inh TO 100 OF exc WITH gaba

List 21.

The first row of list 21 can be configured to generate 800 units of exc types.Second row of list 21 can be configured Into 200 units of generation inh types.The third line of list 21 can be configured to will have mark ' exc ' with connection type glu The individual cell of (noting, Class Type can be used as mark automatically), which is connected to, has 100 that mark ' exc ' or ' inh ' at random The unit of selection.The fourth line of list 21 can be configured to be connected with the individual cell that connection type gaba will have mark ' inh ' It is connected to 100 randomly selected units with mark ' exc '.

Retina pixel to the cone maps

In some implementations, HLND descriptions can be used for description retina pixel to cone mapping.In general, the cone is thin Born of the same parents or the cone can be or may refer in eye retina be responsible for colour vision photosensory cell.Cone cell can be in Intensive filling during centre is recessed, but be towards the periphery of retina and gradually become sparse.The following provide description retina mapping Some examples of various aspects.

100 × 100 square nets (square grid) of // establishment pixel (pixel) coordinate

square_grid(100,100)TAGGED pixels

// pixel cell (pixel unit) is created at each pixel coordinate

PUT pixels OF pixel_unit

The hexagonal mesh (hexagonal grid) of // establishment cone coordinate

hexagonal_grid(100,100)TAGGED cones

// the S cones will be designated as with its 10% random subset

// and the appropriate S_unit (S_ units) of establishment

SIZE(cones)*0.1OF cones TAGGED S_cones

PUT S_cones OF S_unit

// the M cones will be labeled as with 30% random subset in remaining cone

SIZE(cones)*0.3 OF cones AND NOT S_cones TAGGED M_cones

PUT M_cones OF M_unit

// remaining cone is the L cones

cones AND NOT S_cones AND NOT M_cones TAGGED L_cones

PUT L_cones OF L_unit

// each pixel is connected to the S cone near (nearest) by knot (junction)

CONNECT pixels TO NEAREST S_cones WITH p2S_junction

// equally handle the M cones

CONNECT pixels TO NEAREST M_cones WITH p2M_junction

// equally handle the L cones

CONNECT pixels to NEAREST L_cones WITH p2L_junction

List 22.

Digraph

In some implementations, the appointment of token network subset, which may be configured such that, to be digraph by network representation.Have It may include element to G=(V, A) to figure or i.e. orientation diagram, wherein set V element is referred to alternatively as summit or node, and set A For orderly summit to, be referred to as arc, directed edge or arrow.In HLND, term node can be used for summit, and it is side to connect.

HLND and GUI

In some implementations, HLND may include graphic user interface (GUI).GUI is configured to appropriate syntax User action (for example, order, selection etc.) is translated into HLND sentences.GUI can be configured to update in response to HLND sentences GUI is to show the change of network.GUI can provide the one-to-one mapping between the user action in GUI and HLND sentences.It is such a Feature can allow a user to HLND sentences created particularly by display in response to user action etc. and be set with virtual mode Count network.GUI can reflect the HLND languages being for example input to using GUI text editor module during the figure of network is represented Sentence.

Same or similar information (for example, GUI and HLND sentences) in a variety of formats can be allowed by being somebody's turn to do " one-to-one mapping " Unambiguously represented, because different forms is as one man updated to reflect that the change in network design.This exploitation method can It is referred to as " Round-trip Engineering design ".

User action in HLND

In some implementations, GUI can support user action, and it includes creating node, selects the one or more of network Subset, connecting node, the node, and/or other user actions marked in selection subsets.In some implementations, GUI can be supported The definition of network object.Some example users action described in detail below.

Create node

Referring now to Figure 12, illustrate and created according to the use GUI of one or more realizations node.In some realizations In, the node create type that the node of neutral net may require including the node that instantiate and/or generate, to be created The information of quantity, and/or other information.In some implementations, user can be provided including to be assigned to created node The information of list mark, the additional parameter for instantiating and/or initializing node, and/or other information.For instantiating , for example, will on how to be laid out and/or the additional parameter of initialization node can depend on specific real-time performance The instruction of the node (how assigning numerical space to mark) of instantiation.

Above GUI nodes create feature can by HLND kernels realize node generation one or more appropriate instructions To support.More details are see, for example, list above 10.When user's input HLND nodes generation instruction (sentence), GUI can be with In graphic editor generation represented with the accordingly corresponding figure of (all) nodes (for example, uniqueness symbol, picto-diagram, and/or Icon).Input of the user to HLND sentences can be carried out by various means, be included but is not limited to, text input, voice, soft Key (icon), and/or the other means for being configured to HLND input by sentence.

User can perform node establishment using Figure 12 GUI 1200.According to some realizations, user can be from node Selection node type in the freelist (list 1210 in Figure 12) of type；By selected node type (for example, in Figure 12 Class1 216) drag and drop (as explained via the arrow 1204 in Figure 12) enter in editing machine panel 1202, and wherein the node is used Uniqueness node symbol 1208 is represented；(such as illustrated via the supplement input medium associated with specific node Class1 216 For, the popup menu 1220 in Figure 12) additional parameter (if desired) is provided；And/or perform for creating the other of node Action.

Search box 1242 can allow user and filtered using one or more keywords shown node type 1212, 1214 and 1216 list.This can facilitate the node type selection that there are in the case of great deal of nodes type can use.Popup menu 1220 can allow a user to graphically specify the quantity 1226 of node, the parameter 1224 for node instance, layout process 1230th, additional marking 1232, and/or the other information associated with node establishment.

GUI can allow user to be toggled between text editor (HLND sentences 1240) and GUI nodes are created.By By non-limiting example, select that for different parameters/option of node layout HLND sentences can be updated in the gui.Change refers to This information in GUI can be updated by tasking the additional marking of the node created in HLND sentences.

The gui interface shown in Figure 12 is not intended as composition and limited, because other realizations are contemplated and fallen in the disclosure In the range of.For example, in some implementations, GUI may include drop-down list, radio button, and/or other elements.

Select network subset

Referring now to Figure 13,13A and 13B, the different exemplary realizations of Node subsets selection are shown and are described in detail.Figure 13 GUI 1300 may include network topology panel 1302, two or more selection description panels 1304 and 1306, and/or its Its component.The network shown in panel 1302 may include to have is depicted as ' ', ' Δ ', ' zero ' not isolabeling respectively 1305th, 1308 and 1310 multiple nodes.Selection description panel 1304 and 1306 may include the boolean part of HLND sentences, and its is right Should be in respective subset.

In some implementations, " selection network subset " user action may correspond to using GUI editing machines (for example, Figure 13 GUI) member of network is selected.User can (or other pointing devices, such as trace ball, class be touched for example by using mouse Touch finger, light pen, and/or other technologies on pad equipment and iPad from Apple) select the subset of network.Use GUI Subset selection can via optionally click/tapping graphical symbol corresponding with the expectation member of network, click on and drag Drag with select the region of network, its combination, and/or for selecting other actions of subset to reach.GUI subset selection action It can be supported by the command adapted thereto for realizing subset selection of HLND kernels.More details are see, for example, list above 6-7.

As shown in Figure 13, Node subsets 1312 may include the node for including mark 1305, and subset 1314 may include tool There are both nodes of mark 1308 and 1310.Once subset 1312 and 1314 is selected, boolean's table in panel 1304 and 1306 It can be updated accordingly up to formula.

In some implementations, the network shown in Figure 13 GUI 1320 includes two subsets, and it is included comprising mark 1305 and 1308 subset 1322, subset 1314, and/or other subsets comprising mark 1310.Selection description panel 1324 can It is updated to reflect that the Boolean expression corresponding with the mark content of subset 1322.In some implementations, can be by being formed Subset 1326 including the common factor (as indicated by Boolean expression 1328) between subset 1322 and 1314 generates additional son Collection.

By non-limiting example, the Boolean expression for subset case statement is inputted (for example, Figure 13 in response to user In expression formula 1306), GUI can be shown in graphic editor (for example, being shown using the shaded rectangle in Figure 13) should The correspondence of subset selectes member.In some implementations, GUI can generate corresponding with the selection of (all) subsets in graphic editor Figure represent, as explained below in relation to Figure 13 A-13B.Figured example may include uniqueness symbol, picto-diagram, Icon, and/or other figures one or more of are represented.In some implementations, figure represents to may include the change of graphic attribute Change the change of color, shadow mode, and/or other graphic attributes one or more of (including).

Figure 13 A illustrate the exemplary realization of Node subsets selection, and it can be applied to network for including great deal of nodes Collection, wherein individual nodes are described and not always geared to actual circumstances.Figure 13 A GUI realizes that the network shown in 1330 may include Two subsets 1332 and 1334, the rectangle of different shades can be used to describe for it.Figure 13 A GUI realizes the net shown in 1340 Network may include subset, and it can be described by the shape with different filling patterns (see, for example, the He of rectangle 1342 in Figure 13 A 1344).Subset 1346 in Figure 13 A can be selected as 1342 I 1344.

In some implementations, uniqueness symbol 1362 and 1368 can be used to represent for GUI user actions, such as in Figure 13 B Explained in the network 1360 shown.Uniqueness symbol 1362 and 1368 can represent subset 1304 and 1306 respectively, and Can be factor set and different.By non-limiting example, the color of uniqueness symbol and/or other mark quality can according to Configured in the mark of mark subset.The shape of symbol in graphic editor panel 1302 and position can be according to the members of subset Free token configure.This can be explained by symbol 1362,1368,1372,1376,1378,1382 and 1388, and it can be retouched Paint the subset that GUI realizes the network of 1360,1370 and 1380 explanations.In some implementations, symbol can be matched somebody with somebody based on element type Put.In some implementations, symbol choose may depend on the subset include (i) only node, (ii) only connect or (iii) save Point and connection.

In some implementations, identical network configuration (for example, Figure 13 subset 1312) can use distinct symbols/icon Represented in GUI graphics panels (for example, Figure 13 panel 1302).It can correspond to low degree details network view (correspondence In for example without scaling or low-shrinkage put) some realization in, subset can use symbol (Figure 13 B symbol 1362) to represent, and The individual element of volume of the subset need not be shown.

In some realities of the network that may correspond to for example with limited processing capacity or the network for being configured to batch updating In existing, subset can use symbol to represent the individual element of volume without showing the subset.

Can be associated with high level details network view some realization in (correspond to for example high zoom degree, and/or When there is process resource to can be used for handling the information relevant with the individual element of volume of subset), the subset, which can be used, provides the subset The figure of further detail below describe representing (for example, Figure 13 by individual subset elements explanation its appropriate position expression 1312)。

In some implementations, HLND frameworks selection operation can be performed so that additional marking to be assigned to selected member, Connect in sentence using selected member's (being in this case node) and/or the other actions of execution.

GUI can allow user to be toggled between text editor (HLND sentences) and the selection of GUI subsets.By non- Limited example, uses GUI to select different node members can be so that updating corresponding HLND sentences.Change text editing Selection in device can update the selection in graphic editor.In some implementations, update selection may include to be highlighted and/or Separately visually emphasize selected member.

Connecting node

" node connection " user action may correspond to create the connection between network node.According to some realizations, creating When being connected between node, HLND kernels may require it is following one or more：First subset selection (for example, connection by from section Point subset), yield in the second subset selection (for example, connection will end in Node subsets), for the first subset to be connected into the second son The connection type of collection, and/or the other information associated with connecting node.

In some implementations, one or more additional parameters can be provided to HLND kernels, including it is following one or more：With In set connective mapping (for example, it is all to it is all, one-to-one, a pair near, and/or other connections being defined by the user Mapping) parameter, the parameter for instantiating and/or initializing connection example (for example, initialization synapse weight), to be assigned to The list mark, and/or other parameters of the connection example created.

HLND kernels can realize the instruction for being configured to connect the node in network.(more details are see, for example, with above-listed Table 16-17).By non-limiting example, when user inputs link order, GUI can create correspondence in graphic editor Figure (for example, drawing link/arrow of the selection of node from source to destination) is represented to explain these connections.

Realized according to one or more, user can use GUI to select the source subset selection of network memberses, select network The destination selection of member, source selection is dragged and dropped into the selection of destination first choice is connected into the second selection, and/or hold The other actions of row.GUI can generate the link/arrow for representing the corresponding connection between source member and target members in graphics view Head element.

In some implementations, the popup menu associated with connection element (link/arrow) can allow user from connection Connection type is selected in the freelist of type.In some implementations, popup menu can allow user to provide for instantiating And/or the additional parameter of initialization connection example.In some implementations, popup menu can allow user to be provided for the company of setting The parameter of general character mapping.

GUI can allow user to be toggled between text editor (HLND sentences) is connected establishment with GUI.By non- Limited example, uses GUI to select different node members to update the HLND sentence associated with node description. The selection changed in text editor can update the selection (for example, being highlighted selected member) in graphic editor.

Figure 14 illustrates using GUI to connect the exemplary realization of two node sets.GUI 1400 may include network section Point view panel 1402, one or more node sets selection domain 1406 and 1404, HLND sentences domain 1442 and 1444, and/or Other components.When being clicked in HLND GUI domains 1402 using such as mouse and drag kick selects the He of node set 1412 When 1414, node set selection domain 1404 and 1406 can be updated to reflect that selected set.As explained in Figure 14, set 1412 may include the node with mark 1405 (being depicted as ' square ').Set 1414 may include (to describe with mark 1408 For ' triangle ') and node with mark 1410 (being depicted as ' circle ').Node set selection domain 1404 and 1406 can be by more It is new to show selected node with 1412 and 1414 corresponding marks of set respectively.

According to some realizations, new mark can be assigned for the selected member of network by a selection up-regulation with such as right click Note is assigned additional marking to set 1412 and 1414.For marking the HDLN sentences of addition to be automatically generated.

In some implementations, such as right click (as line 1418 is explained) can be used to call supplement graph data input hand Section (for example, popup menu 1430 in Figure 14).Menu 1430 can be used for additional marking 1432 or new mark 1434 especially It is assigned to selected node etc..

In some implementations, it is dynamic by using drag and drop as the arrow 1416 in Figure 14 in GUI domains 1402 is explained Make, using the finger on mouse and/or touchpad devices, first choice ' can be put ' onto the second selection, and this can instruct HLND Engine creates the connection between the node 1405 of set 1412 and the node 1408 and 1410 of set 1414.

Additional supplement graph data, which keys in means (for example, popup menu 1420 in Figure 14), can be particularly useful for specifying Parameter for connection.Specify for the parameter of connection may include it is following one or more：Connection type 1422, initialization are set Parameter 1424 for connection type, specify connection sexual norm 1426, mark is assigned to connection 1428, and/or other actions.

It is corresponding for selecting the 1412 and 1414 one or more user actions performed with node using GUI 1400 HLND sentences can be automatically generated and are shown respectively in sentence domain 1442 and 1444.

The gui interface shown in Figure 12-14 is not intended as composition and limited, because other realizations are contemplated and fallen in the disclosure In the range of.For example, some realizations may include drop-down list, radio button, and/or other components.

HLND and END forms relation

HLND forms are designed to describe (END) format compatible and/or in connection, END with element formula network Entitled " the ELEMENTARY NETWORK DESCRIPTION FOR submitted for 21st in September in 2011 that form is included more than NEUROMORPHIC SYSTEMS (the element formula network for being used for neuron morphology system is described) " U.S. Patent application S/ N.13/239,123 described in.In some implementations, can be described based on HLND the END of model neuron (for example, realize) come Generate the example of END units.END is tied and/or the example of END cynapses can directionally connect each unit.HLND can define dissection Structure, and nerve and cynapse dynamically can be defined in the END classes applied.HLND can hide END complexity and/or rudimentary Difficulty, and network design can be made to be simple procedure.

The END examples generated can be used for the neutral net engine that generation is realized and/or runs designated model.That is, END Example can be used for the engine that the network that the END classes for being described and/or being applied by HLND are defined is realized in generation.The engine can be Performed on PC, FPGA, any hardware platform of any special END compatible hardwares, and/or other computer hardwares.

Figure 11 illustrates the three basic structures of END engines, its can general RISC/CISC CPU (CPU), Realized on graphics processing unit (GPU), integrated circuit (for example, ASIC), and/or other processors.The structure of END engines can be with Corresponding to ' unit ' 1101 in Figure 11, ' doublet ' 1111, and/or ' triplet ' 1121.END engines can be configured to perform Unit, doublet and triplet rule, and/or the memory for accessing these elements.END forms can be taken as to have configuration and hold Row specifies the hardware specification language of the semiconductor circuit of the such unit, doublet and triplet of neuroid to treat.

In some implementations, individual basic structure (for example, unit, doublet, and/or triplet) can be implemented as multi-thread Single thread on thread processor.In some implementations, individual configurations can be implemented as hyperelement, super doublet, and/or surpass three Conjuncted, it may include to be configured to the special circuit that use time multiplexing respectively comes processing unit, doublet, and/or triplet.One A little realize may include three different circuits：Each of unit, doublet and triplet have a circuit.

In some implementations, unit 1101 can represent the part (for example, dendron chamber) of neuron and/or neuron. In another example, unit 1101 can represent the colony of neuron.The activity of neuron can represent the " average to swash of the colony Hair rate " activity and/or movable other mean field approximations of the colony.Individual cell can be with memory variable and description to it The renewal rule what operation memory can perform is associated.Each operation can be (that is, in each of simulation based on clock Time step perform) or they can (that is, be performed based on event when some events are triggered).

Depending on the value of element variable, these units can generate via doublet to trigger the cynapse thing in other units The event (for example, pulse or spike) of part.For example, the unit 1102 in Figure 11 can influence unit via doublet 1111 1103, doublet 1111 can represent that (postsynaptic is single from presynaptic neuron (cynapse front unit 1102) to postsynaptic neuron The cynapse of member 1103).

Individual cell, which can have, updates rule after event, the rule is triggered after being triggered in event.This A little rules can be responsible for element variable and be changed caused by event, for example, being reset after the spike of voltage quantities.

The doublet of individual can be associated with memory variable.The doublet of individual can access the change of postsynaptic unit Amount.Such a access may include reading and writing, and/or access mechanism.The doublet of individual can change with being made to duplex body memory Become is associated with the doublet event rules for realizing synaptic plasticity.Individual doublet can with to postsynaptic cell memory Make changing is associated with the doublet event rules for realizing pulse delivering.Doublet event rules can cover the above in END Some or all of cynapses rule described in form.

Because multiple doublets (for example, 1116-1118 in Figure 11) can be by corresponding multiple cynapse front unit 1106- 1108 are connected to single postsynaptic unit 1109, so doublet can change postsynaptic list parallel and/or with any order Metamemory.As a result can be unrelated with order.This can be in the operation to postsynaptic cell memory atom addition (such as In GPU like that), atom multiplication (it is equal to the addition via logarithmic transformation), and/or to reset to a value (wherein all double Conjuncted trial resets to same value) when reach.The postsynaptic element variable just changed by doublet event rules cannot be at this Used in rule.Otherwise, as a result it is likely to be dependent on the execution order of doublet event rules.

Referring now to Figure 15, show and be described in detail to define system including HLND kernels and the END neutral net described Exemplary realization.In fig .15, circle 1504,1502 and 1506 can represent that different relatively high-level networks describes method or lattice Formula.Circle 1510 can represent the END descriptions of network.END descriptions can be indicated to from 1504,1502 and 1506 arrow Transfer process.For example, the END that the software of the HLND descriptions (for example, HLND sentences) of processing network can generate consolidated network is retouched State.Rectangle 1512,1515,1516 and 1518 in Figure 15 can indicate the various hardware for the network that 1510 definition are described by END Platform is realized.Arrow between circle 1510 and rectangle 1512,1515,1516 and 1518 can indicate engine generating process.By By non-limiting example, the arrow that END is described between 1510 and rectangle 1512 can represent that generation is realized END networks and is configured Into the process for the executable item run on CPU.HLND definition 1502, which can be processed and be converted to END, describes 1510.END is described It can be configured to be processed (for example, being handled by each software application) to generate the machine-executable instruction different because of platform.This A little machine-executable instructions different because of platform can be configured to perform on a variety of hardware platforms, these hardware platforms include but It is not limited to element general processor 1512, graphics processing unit 1514, ASIC 1516, FPGA 1518, and/or other hardware flat Platform.

Other network descriptor formats can be combined with process 1500, for example, BRIAN 1504 and/or be configured to life In the other neuron morphology network descriptor formats 1506 (for example, NEURON (neuron)) described into the END of network, such as Figure 15 Explained.

The exemplary realization of the HLND devices of computerization

Had shown and described on Figure 16 and be configured to utilize in design neutral net (for example, Figure 15 network 1500) The exemplary realization of the computerization network processing device of HLND frameworks.Computerized device 1600 may include process block (for example, Processor) 1602, it is coupled to non-volatile memory device 1606, random access memory (RAM) 1608, user's input/defeated Outgoing interface 1610, and/or other components.User's input/output interface may include it is following one or more：Keyboard/mouse, figure Display, touch-screen input-output equipment, and/or it is configured to receive input and/or to the other of user's output information from user Component.

In some implementations, computerized device 1600 can be via (such as, the computer I/O buses of I/O interfaces 1620 (PCI-E), wired (for example, Ethernet) or wireless (for example, WiFi) network connection) coupled to one or more external treatments/ Storage device.

In some implementations, input/output interface may include to be configured to set from the phonetic entry that user receives voice order Standby (for example, microphone).Input/output interface may include to be configured to the speech recognition mould that voice order is received and identified from user Block.The various methods of speech recognition are considered as to fall within the scope of this disclosure.The example of speech recognition may include following one Or many persons：The spectral analysis algorithm including Mel Cepstral Frequency Coefficients based on linear predictive coding (LPC) run on a processor (MFCC) spectrum analysis, cochlea modeling, and/or other methods for speech recognition.Phoneme/word identification can be based on HMM (hidden markov modeling), DTW (dynamic time warpage), NN (neutral net), and/or other processes.

END engines 1510 be can be configured to the HLND descriptions of network being converted to machine executable format, and the machine can be held Row format can be realized for specific hardware or software and is optimised.Machine executable format may include can be by process block The 1602 multiple machine-executable instructions performed.

Skilled artisans will appreciate that, various processing equipments can be combined with various realize, including but not limited to monokaryon/many Core CPU, DSP, FPGA, GUP, ASIC, its combination, and/or other processors.Various user's input/output interfaces can be applied to It is various realize, include but is not limited to, LCD/LED monitors, touch-screen input and display device, voice-input device, instruction pen, Light pen, trace ball, and/or other user interfaces.

The execution of GUI user actions

In some implementations, network design system (for example, Figure 16 system 1600) automatically can be converted to GUI action HLND instructs and/or is converted into END statement.HLND instructions may cause to automatically update GUI and represent and/or END descriptions.

Figure 17 illustrates a kind of method for the seamless renewal for performing the different expressions corresponding from consolidated network design element. Network describes 1702 (for example, node, connection, subsets etc.) can be comprising information necessary to defining network.In some implementations, net Network description (1702) may include one in node type, node type parameter, node layout's parameter, mark, and/or other information Person or many persons.In some implementations, network description (1702) may include connection type, connection type parameter, connection mode, mark One or more of note, and/or other information.There may be it is various other description types (for example, subset), its may include with Associated just suitable information.

As explained in Figure 17, single object (such as object 1702) can have one or more tables associated therewith Show, these expressions may include GUI represent 1712 (for example, using Figure 13 GUI editing machines 1302), HLND represent 1714 (for example, Use the HLND sentences described above with respect to Figure 14), END represent 1716 (see, for example, Figure 15), and/or it is other expression (by square Shape 1718 is described).In response to the object property (that is, object data element) for being just generated and/or updating, the correspondence of the object Represent that (for example, representing 1712,1714,1716 and 1718) can be respectively using two-way approach 1720,1722,1724 and 1726 Update.

In some implementations, in response to user's GUI action of the selection of modification one, corresponding (all) HLND sentences are (for example, figure HLND in 17 represents 1714) to be updated.In some implementations, END instructs (for example, the END in Figure 17 represents 1722) can It is updated.

In some implementations, END instructions can be performed by device, can thereby realize to network specifically and accurately Expression.

In some implementations, when the sentence of creating unit can use in network describing framework, uniqueness can be used color Node is presented in color symbol in GUI.

In some implementations, can use in response to the coordinate of node --- i.e. when connection sentence is handled at least in part When, each node can be presented on its proper position in GUI with the symbol of uniqueness (for whole subset).

In some implementations, when link order in network describing framework can use when, can by GUI using such as single line Lai The connection between two subsets is presented.

In some implementations, once information is available (that is, previously raw before the node for connecting example and after node Into) --- i.e., once connection sentence is handled at least in part, then the connection between two subsets can just show to connect in detail Connectivity structure.

In some implementations, in response to information, and/or initial weight after information, node before the node for connecting example Can use --- i.e., once connection sentence is handled at least in part, then the connection between two subsets just can be with often connecting The property line width of connection (for example, represent) of uniqueness shows structure connective in detail.

It is other to represent that (for example, what the rectangle 1718 in Figure 17 was described) as understood as those skilled in the art With exist and can be compatible with various realizations, on condition that they meet more new frame described herein.

In some implementations, the different numbers represented between (for example, expression 1712,1714,1716 and 1718 in Figure 17) According to exchange can via the direct link indicated by arrow 1730,1732,1734,1736,1738 and 1739 in Figure 17 come Realize.For the sake of clarity, do not show that between expression 1712,1714,1716 and 1718 in Figure 17 all is directly connected to.

It will be recognized that the specific steps order while in accordance with method describes some aspects of the disclosure, but these are described The wide method of the disclosure is only illustrated, and can be modified as needed to application-specific.In some cases, it can cause Some steps are unnecessary or can be optional.In addition, some steps or feature can be added into disclosed realization, or two Or more the execution order of step can replace.All such variants are considered as to covered in disclosed herein and require In the disclosure of protection.

Although the disclosure is described in detail for purpose is explained based on the most realistic and preferred realization being presently believed to, It will be understood that, such details is only used for the purpose and the disclosure is not limited to disclosed realization, but on the contrary, the disclosure is intended to contain Lid falls modification and equivalent arrangements in the spirit and scope of the appended claims.For example, it will be appreciated that the disclosure is as far as possible One or more features of any realization are contemplated in degree to be combined with one or more features of any other realization.

Claims (15)

1. it is a kind of generated in the neutral net including multiple elements multiple connections by computer implemented method, methods described Including：

Performing at least includes the first logical expression of the first mark and the second mark；

The execution is based at least partially on, the first subset and yield in the second subset of the multiple element is identified；And

Generate multiple connections between at least a portion of first subset and at least a portion of the yield in the second subset；

Wherein：

Each element of first subset includes the described first mark；And

Each element of the yield in the second subset includes the described second mark.

2. the method as described in claim 1, it is characterised in that every unitary of first subset and the yield in the second subset Element includes the node of the network.

3. the method as described in claim 1, it is characterised in that further comprise the described first mark being assigned to described first Each element of subset.

4. the method as described in claim 1, it is characterised in that at least one of described first mark and the described second mark Characterized by limited life cycle.

5. the method as described in claim 1, it is characterised in that each connection in the multiple connection includes cynapse and knot One.

6. method as claimed in claim 2, it is characterised in that mark first subset is configured so that and can generated The new network element of at least a portion node including first subset.

7. a kind of processing unit, including the non-volatile memory medium of multiple instruction is configured to store, the multiple instruction exists Carry out and the dynamic of neutral net is divided when being performed according to a kind of method, methods described includes：

Identify the subset of elements of the neutral net；And

The each element for being assigned to the subset of elements will be marked, the mark includes the mark for being configured to identify each element Know symbol；

The wherein described appointment mark, which is configured so as to generate, includes the new network element of the subset of elements.

8. device as claimed in claim 7, it is characterised in that methods described is used using application specific integrated circuit (ASIC) ASIC instruction set is realized.

9. device as claimed in claim 7, it is characterised in that methods described further comprises performing quilt by the processing unit It is configured to identify the mathematic(al) representation of each element of the subset.

10. device as claimed in claim 9, it is characterised in that the mathematic(al) representation includes Boolean calculation.

11. processing unit as claimed in claim 10, it is characterised in that each element of the subset is using random Selection operation carrys out selection.

12. processing unit as claimed in claim 7, it is characterised in that methods described further comprises assigning the mark To the new network element.

13. device as claimed in claim 7, it is characterised in that described the mark is assigned into the subset to be configured to It is digraph to make it possible to the network representation.

14. device as claimed in claim 7, it is characterised in that methods described further comprises the second mark being assigned to institute Subset is stated, second mark and the mark are different.

15. a kind of system for being configured to generate multiple connections in the neutral net including multiple elements, the system includes：

It is configured to perform the one or more processors of computer program module, wherein the execution of the computer program module makes One or more of processors：

Performing at least includes the first logical expression of the first mark and the second mark；

Perform to identify the first subset and yield in the second subset of the multiple element, wherein institute based on first logical expression State the first subset individual element of volume include described first mark and individual element of volume of the yield in the second subset include described second mark； And

Generate multiple connections between at least a portion of first subset and at least a portion of the yield in the second subset.