Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology

Definitions

• Computing system for simulating the cerebral cortex.
• the invention relates to a computer system for simulating the cerebral cortex with the radial glia, i.e. a relation computer according to the preamble of claim 1.
• DE-PS 34 29 078 of the applicant specifies a computing system for simulating the format io reticularis, with which e.g. a robotic system executes certain intended actions specified by programs with redundancy of potential command execution.
• the intended actions are predetermined by framework programs, which, however, are determined by detailed, e.g. data derived from environmental information can be re-evaluated and also exchanged, but the goal of the intended action must be kept in mind. How the path to this goal proceeds is determined by the structure of the entire computer system and the logic used within the system.
• the relationship calculator is an essential part of such an overall computing system.
• the data calculated from the environmental information and the intended actions are input into this relationship computer, with feedback from the command computer being present at the same time. These input data are weighted accordingly in the relation computer and finally passed on to the command computer.
• this relation computer corresponds to the cerebral cortex of the biological brain; see. WL Kilmer et al. in International Journal of Man-Machine Studies, 1969, volume 1, pages 279 to 309, in particular FIG. 10 on page 306 with the associated description.
• the relationship computer which consists of a large number of classic computers for processing various environmental information, deductive motor programs and inductive planning, is not specified in the prior art only with regard to its function however with regard to the computer structure.
• the cerebral cortex consists of a large number of (ontogenetic) columns, each of which contains a certain number of neurons.
• the structure of these columns starts from a layer of glial cells, the radial glia, which no longer belongs to the cerebral cortex, but to the ventricular zone.
• Each glia cell arranged in the ventricular layer is connected to a column of the cerebral cortex via radial glia fibers, there being a unambiguous association between the glia cells of the ventricular zone and the columns of the cerebral cortex.
• a certain cerebral cortex area is built up under the guidance of the glia cells.
• the neurons required for the individual columns migrate along the radial glia fibers or extensions of the glia cells to the columns of the cerebral cortex and there exactly to the corresponding destination within the column.
• the final number of such pillars of each area can be modified by afferent information.
• the cytoarchitectonics of the cerebral cortex, as observed in practice, are hereby divided into function-specific areas.
• the cerebral cortex enlarges as the number of radial glial cells increases. It is particularly noteworthy that the enlargement of the surface of the cerebral cortex in the course of evolution is not accompanied by a substantial increase in the thickness of the cortex.
• the invention is based on the object of converting this knowledge gained from biology into an indication for a computer structure that can be used to simulate the function of the cerebral cortex with the other parts of the brain participating in it.
• the columns of the cerebral cortex correspond to computer groups and the glial cells of the ventricular layer cells of a tritometer, each of which is assigned a specific tritogram.
• the cells within a trito-counter are organized in a kenogram and are arranged according to their Deutero equivalence.
• the tritograms and the Deutero equivalence cf. DE-OS 37 07 998, G.G. Thomas, Introduction to Kenogrammatics, Proceedings of the 13th Winter School on Abstract Analysis Section of Topology, Serie II, No. 11, 1985, and G. Günther, Logic, Zeitemanation und Evolution, Geistesritte Hefte 136, Obladen, West Germanyr Verlag 1967, pages 7 to 47.
• the kenogrammatically organized and deuterographically ordered cells of a Trito counter are thus uniquely assigned to computer groups, also arranged in groups.
• Each computer group receives particular group-specific data, such as the data of a perception computer, which provides visual or tactile environmental information, the data of an action intention computer, the one to be executed Provides programs, or feedback data from a command computer, etc. If calculation is carried out in a certain computer group of the relationship computer, it being irrelevant how many individual computers are active, an activity switch assigned to the computer group is switched on, which only sends the via a line to the associated Trito cell Provides information, for example the information ON, that calculations are carried out in the associated computer group.
• the cells of the tritone counters that are switched on are queried at certain time intervals, in which case the tritogram of the cells that are turned on is then forwarded to a memory, for example a RAM module, as a so-called qualitative pattern.
• the qualitative patterns of the current computing processes are hereby continuously stored in the relationship computer, whereby the current quality pattern can be made accessible to an observer at any time, for example via a monitor. Since the tritograms correspond to value qualities, these value qualities are also the basis of the action intention program, a comparison of the two quality or program samples can be used to determine whether the action program is being changed, for example, according to a certain environmental experience, which is expressed in a new quality sample of the tritograms must to achieve the desired goal.
• the computer structure or computer architecture presented is a simulation of the interaction of glial brain structures with neuronal brain structures. It is therefore not just a so-called neural computer, but a glia-neuronal computer system.
• the structure of this computer system according to the kenogrammatics shows a corresponding increase in the number of assigned computer groups even with an increase in the tritocells in the course of further computer development, without the expansion of the individual computer groups becoming significantly larger, which is what the Fact observed over the course of evolution,
• the basic structure of the relationship calculator is also transferable to the computer's subsystems.
• relation computer also functions as a perception computer with a sensor system
• the sensor system then consists of individual sensor units, each associated with a tritogram and one corresponding to the symbols or kenograms of the assigned tritogram
• Cerebral cortex corresponding computer groups summarized in Deutero-equivalent sensor groups.
• the value assignments of the individual tritograms, each of which corresponds to a certain number of individual computers in the above computer group, are used in the sensor system as a code for a corresponding sensor.
• the possible and permissible values of the individual places within a tritogram then correspond to a certain information quality
• Figure 1 is a schematic partial perspective view of part of a cerebral cortex and related glia cells in the ventricular zone in the biological brain (after P. Rakic).
• 2 shows a block circuit diagram of a computing system containing an action intention computer, a command computer and a relation computer according to the invention
• Fig. 3 is a diagram of the connection between a part
• Fig. 4 is a graph for explaining the development
• Fig. 5 is a schematic representation of a sensor system
• Fig. 6 is a tritogrammatically ordered group of computers
• FIG. 7 shows a schematic illustration of a permutograph computer as a pattern recognition computer in the perception computer
• FIG. 8 shows a diagram of a perception process as pattern recognition with a perception computer organized in the manner of the relationship computer.
• Fig. 1 shows schematically some glia cells GC in the ventricular zone VZ, the radially extending
• CP are connected.
• neurons move along the glia that serves as a guide Fibers RG, so-called migrating neurons MN (migrating neurons) in the direction of the assigned columns C, migrate past neurons which have already been deposited in the region between E40 and E100 and are then deposited in the direction of the edge region MZ.
• migrating neurons MN migrating neurons
• Each glia cell GC is thus assigned a specific column C.
• the neurons in columns C receive information about connections to the interbrain TR and about connections to other areas of the cerebral cortex CC. The representation and explanation of this figure is - borrowed from the above-mentioned article by P. Rakic.
• the following table 2 shows the number of possible tritograms for the number of n symbols leads, this number corresponds to the sequence of the hatchet numbers B (n).
• the possible value assignments are then (1st ) (1) (1) (1), (2) (2) (2) (2), (3) (3) (3) (3) and (4) (4) (4) (4).
• the tritogram T2: 1 1 1 2 listed in the second column of Table 1 therefore means that in the case of four values, a sequence of three identical values and a different value each. Accordingly, 12 different value assignments can be set up, starting from (1) (1) (1) (2), (1) (1) (1) (3) until finally (4) (4) (4) (3) , thus a total of 12 value assignments.
• this explanation means that in each case the column which is assigned to a specific glia cell is assigned a number of neurons which is equal to the number of value assignments which corresponds to the tritogram of the respective glia cell.
• tritograms can each be combined to form kenographs, the structure of the kenograph being determined by the deutero equivalence of the tritograms.
• Each tritogram can be represented as a Deuterograram, whereby only the distribution of the different symbols is relevant for this representation.
• tritograms in their deuteroequi va lent structure can each represent coherent cenographs, as already explained in the above-mentioned DE-OS 37 07 998.
• the tritograms T2, T3, T6 and T10, furthermore T4, T7, T9 and T5, T8, T11, T12, T13, T14 can then be combined in groups, whereas the tritograms T1 and T15 remain solitary.
• This group summary can be understood as qualitative functional equivalence.
• This computing system has an action intention computer, a command computer 2 constructed as a permutograph and a relation computer 3, the function of which is explained in DE-PS 34 29 078, to which reference is made.
• the relationship computer 3 in question here has a tritometer counter 4 with fifteen cells Z1 to Z15, computer groups 5 represented as columns S1 to S15, in each of which a computer location is represented by a small circle, a sample memory 6 connected to the tritometer counter 4, one frequency counter 7 connected to this and a monitor 8 connected to the pattern memory.
• a tritogram is assigned to each cell Z1 to Z15 of the tritometer, the cells with deuteroequivalent tritograms being arranged adjacent to one another, which is indicated by the double dashes. It can be seen that cells Z1 and Z15 are singular with the tritograms T1 and T15 indicated only by square boxes, and the other cells are arranged according to Table 3 above. Cells Z2, Z3, Z6 and Z10 form a group, as do cells Z4, Z7 and Z9 and cells Z5, Z8, Z11, Z12, Z13 and Z14.
• the Trito Counter 4 is therefore a technical one Equivalent to the arrangement of the glia cells GC in FIG. 1.
• the tritometer is a counting device based on tritograms, tritograms in turn counting or reducing a certain quantity of values to a corresponding value quality.
• Each cell Zi of the tritometer 4 with its tritogram is assigned exactly one computer group Si, each of which has a number of computer locations corresponding to the tritogram of the cell, in accordance with the value assignment explained above. Accordingly, the computer group S1 has four computer locations, the computer groups S2, S3, S6, S10, S4, S7 and S9 each have twelve computer locations, while the other computer groups S5, S8 and S11 to S15 each have twenty-four computer locations.
• These computer workstations are each available in the same quality as classic computers, which handle data and information supplied by other computer systems in accordance with their tasks, as explained below.
• the unambiguous assignment between the cells Zi and the computer groups Si creates subspaces made up of specific columns within a column space formed by a large number of adjacent columns, which belong together qualitatively. This togetherness is also indicated by double dashes. This togetherness corresponds to a task-specific subdivision of the computer function in accordance with a specific task-specific subdivision within an area of the cerebral cortex.
• the computer workstations available in a computer group Si do not all have to be occupied by one computer, since the installation of the required computer capacity depends on the particular tasks that the relation computer has to perform. Even computers that have already been installed can be taken out of operation again. which would correspond to the actually observed biological decline of neurons in the cerebral cortex.
• an activity switch 9 connected to this computer group is switched on, which delivers a signal to the respectively assigned cell Zi in the tritometer counter 4.
• the tritograms of those cells Zi which each receive an ON signal from an activity switch and are read out in the pattern memory 6, e.g. a RAM block. This stored pattern can be made accessible at any time via the monitor 8.
• the tritogram patterns stored in the pattern memory 6 are reported to the computer groups S1 to S15 via lines 10.
• a line 10 is e.g. a feed forward line, i.e. it transmits data when certain Uraweltinformation meets a quality pattern currently existing in the computer system, thus corresponds to an affirmation in the sense of an emotion, or it is a feedback line that only serves to report back the quality pattern.
• a feed forward line i.e. it transmits data when certain Uraweltinformation meets a quality pattern currently existing in the computer system, thus corresponds to an affirmation in the sense of an emotion, or it is a feedback line that only serves to report back the quality pattern.
• only the feed-forward line is necessary for a simulation of computing processes in the brain.
• the tritogram patterns that occur are counted and the most frequently occurring patterns are stored in the frequency counter 7 and transmitted to the action intention computer 1.
• the action intention computer 1 can then, if necessary, modify its program from these patterns, which result from an intended action program. Modifications are communicated directly to computer groups S1 to S15 or to command computer 2 via lines 11.
• any computer system can be built according to the architecture of the relationship computer. In the following, the special case of a robot system according to the model of DE-PS 34 29 078 will be explained.
• the computer groups S2, S3, S6 and S10 calculate the environmental information in relation to the sensors, e.g. regarding sight, hearing, touch and smell.
• the computer groups S5, S8, S11, S12, S13 and S14 contain learnable, i.e. Inductive computers for calculating planning and weighting programs according to plans 1 to 6.
• learnable i.e. Inductive computers for calculating planning and weighting programs according to plans 1 to 6.
• the data perceived by the sensors are also offset against the plans intended for action.
• Computers are provided in the computer group S15, which determine the bidding and prohibition logic of the robot system.
• DE-OS 37 07 998 has shown how tritograms can be converted into permutations or permutations into tritograms.
• the conversion of permutations into tritograms is unambiguous, whereas the reverse conversion can also be ambiguous, ie several permutations are assigned to a tritogram.
• This assignment also requires the functional connection between the relationship computer and the command computer designed as a permutograph, as shown in FIG. 3.
• the command computer is shown there from 24 nodes 1 to 24 each represented as circles, which are connected to one another according to a very specific scheme, cf. the aforementioned disclosure.
• Converters 12 are now connected to cells Z1 to Z15 of tritometer counter 4, which convert the tritograms present in the cells into permutations. Depending on which permutation is calculated here, a connection is made to the node assigned to this permutation.
• the conversion of the tritogram T15 of the cell Z15 leads, for example, to the permutation 1 2 3 4, which is assigned to the node computer 1.
• Such a clear conversion of tritograms into permutations also applies to cells Z4, Z7, Z9, Z5, Z8, Z11, Z12, Z13 and Z14, so that the associated cells of the tritometer with the corresponding nodes 8, 17, 24, 7 , 15, 3, 22, 6 and 2 of the command computer 2 are connected.
• the tritograms of the cells Z2 and Z3 each lead to two permutations which are assigned to the node computers 13 and 9 or 20 and 12.
• the tritogram of cell Z1 can be converted into six permutations which are assigned to node computers 23 19, 18, 14, 11 and 10, respectively.
• the cell Z1 is therefore connected to the six corresponding node computers.
• the cells of the tritometer 4, from which lines lead to two or more node computers of the command computer, are accordingly assigned to computers within the computer group ⁇ S1 to S15, which are primarily intended to process the current environmental information and to guarantee the security of the robot system in its field of work.
• Cell Z1 connected to computer group S1 is even assigned six lines to six different node computers. Since this computer group is concerned with the current environmental monitoring, the command computer must be able to be informed about any emergencies redundantly, ie if necessary via several node computers.
• the computer group S1 is, so to speak, an "emergency column" which receives its information directly from the computers of the environmental sensors.
• the technical implementation of the relationship calculator was based on the model of the biological cerebral cortex, including the kenogram, together with the tritograms.
• the inclusion of kenograms has another surprising consequence.
• the brain develops, essentially only the area of the cerebral cortex and the associated parts increases, but not their thickness.
• 4 shows the formal-structural structure of tritograms, a so-called tritostructure, which develops from the tritogram symbol 1 (according to G. Günther, loc. Cit., Page 24).
• FIG. 5 schematically shows a sensor system 51 for simulating the tactile sensor system, ie sensor system that responds to touch.
• a four-valued computer system is again assumed.
• 15 tritograms are possible with such a computer system.
• 15 sensor units Se1 to Se15 are provided, each of which has four individual sensors 52 corresponding to the number of symbols in each tritogram.
• the sensor units Se1 to Se15 are combined according to the Deutero equivalence, as explained above for the computer groups S1 to S15, to sensor groups G1, G2, G3, G4 and G5, so that the group G1 is one sensor unit, the group G2 four sensor units, the group G3 three sensor units, the group G4 six sensor units and the group G5 in turn only one sensor unit.
• the group G1 is one sensor unit
• the group G2 four sensor units the group G3 three sensor units
• the group G4 six sensor units and the group G5 in turn only one sensor unit.
• corresponds to each sensor has a place in an assigned tritogram. While the 15 columnar computer groups of the above
• Sensor groups G1 to G5 are combined, fed to a perception computer 54, which is organized like the relationship computer described above. Accordingly includes
• the computing unit of the perception computer 54 according to FIG. 6 again 15 computer groups S1 to S15, which are occupied by individual computers and are arranged among one another in accordance with the Deutero equivalence such that, as above
• the first column S1 which comprises only one computer group, is an on / off switching module 61, in which the temporal effect of information on the complete sensor system is registered in terms of the duration of a touch stimulus.
• the second group of columns (S2 - S3 - S6 - S10) forms a logic computer 62.
• the computer system specified in the unpublished German patent application P 39 33 649.2 (step pyramid computer system) is suitable for four computer groups, so that at least one step pyramid computer is accommodated in each column.
• Two step pyramid calculators compare places and occupancy and use them to calculate the logical functions.
• conventional computers can also be installed in this column group 62.
• the third column group (S4 - S7 - S9) forms an intensity or frequency decoder 63.
• Columns S4, S7 and S9 have four places corresponding to the four individual sensors, for each of which three codes, i.e. Standard values for the tritograms are possible. In this case, two lines lead to two adjacent places on a column, which correspond to two different frequency ranges from the respective sensor.
• the perception computer can distinguish touch frequencies.
• the frequency distribution of the occurrence of a specific frequency range is registered by a frequency range counter and made available in a memory for further comparisons.
• the column group (S5-S8-S11-S12-S13-S14) forms a pattern recognition computer (64), which is shown in more detail in FIG.
• S13 and S14 of the perception computer are networked as a four-value permutograph computer 65 with 24 node computers k1 to K24.
• Each individual sensor 52 of the isotopic sensor system 51 corresponds to a node K in the permugraph computer 65.
• the numbering of the 24 coding units of the columns S5, S8, S11, S12, S13 and S14 indicated in FIG. 5 corresponds to the 24 permutation addresses of the command computer which result from the tritograms developed the individual sensor units Se5, Se8, Se11,
• Se12, Se13 and Se14 are assigned:
• the assigned tritogram is T5 (1 1 2 3, see above), i.e. it is composed of two at first
• Permutation can appear as a numerical value between 1 and 4, so there are 24 possible permutations that
• Sensor group Se5 can be assigned as standard value assignments. In the same way, the codes or
• the last column group S15 forms a pattern generation computer 66, in which perception patterns are generated in all possible contextures provided.
• This computer is also preferably designed as a tetravalent permutograph computer.
• This permutograph calculator is equivalent to that in the German one
• Patent 33 28 610 (U.S. Patent 4,783,741) to an apparatus
• the computer group S15 has
• the perceived patterns sought are transmitted in parallel to the pattern recognition computer 64 in the form of the permutograph computer 65.
• This permutograph computer 65 now places the isotopic sensor unit on the perception pattern in question. Depending on whether the individual sensors 52 are touched, the touch pattern more or less corresponds to the pattern sought.
• a specific pattern P1 sought by the perception computer 54 is compared in a comparator 71 with the pattern P2 actually felt by the sensor system 51.
• This comparison result is fed to a memory logic 72 in which already learned patterns are stored.
• a comparison is made with patterns that have already been learned in a further comparator 73, and on the other hand, in an optimizer 74, a pattern that is only partially recognized in the comparator 71 is optionally supplemented to form a more complete or a complete pattern.
• two decisions are made in the memory logic 72:
• the pattern P1 sought by the perception computer 54 is recognized: Then this pattern is sent to a monitor
• the above-mentioned contextual calculator 76 instructs to change the contextual. This command is then forwarded to the perception computer 54 so that the perception process begins again.
• the instantaneous touch pattern becomes direct from the sensor system to the permutograph computer
• the pattern recognition is therefore controlled twice, firstly via the pattern memory in the memory logic 72 and secondly directly via the touch pattern on the sensor system 51.
• the generation of perception patterns in the pattern generation computer 66 is thus also one of the intended action programs (running in the command computer) dependent. This corresponds to the results
• Command computers can be represented, so there is
• the perception calculator is not in contexts (according to Werner), but in
• Brain cortex are therefore each for a sensor quality
• German patent 36 09 925 described two-tube computer system a suitable system to continuously compare, e.g.

Landscapes

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

Applications Claiming Priority (2)

Publications (1)

Family

ID=6377313

Family Applications (1)

Country Status (4)

Families Citing this family (5)

Family Cites Families (5)

• 1989
  • 1989-03-28 DE DE3910036A patent/DE3910036C1/de not_active Expired - Lifetime
• 1990
  • 1990-03-27 WO PCT/EP1990/000488 patent/WO1990011575A1/de not_active Application Discontinuation
  • 1990-03-27 EP EP90904826A patent/EP0416076A1/de not_active Withdrawn
  • 1990-03-27 US US07/623,963 patent/US5410716A/en not_active Expired - Fee Related

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events