JP2000215264A - Signal processing unit and circuit therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a WTA competition process occurring between synapses in a living body electrochemically. SOLUTION: Plural signal processing units are installed for the same anion solution. Each signal processing unit is formed of a pair of electrodes formed of a first electrode 18a where a polypyrrole layer 16 is formed on a surface and a second electrode 18b where the polypirrole layer 16 is not formed on the surface, and a power source part applying voltage to a pair of electrodes. The power source part switches a first power source part applying voltage in sizes which are individually decided for the respective first electrodes 18a based on inputted data and a second power source part applying uniform inverse voltage to all the second electrodes 18b by a switch mechanism at a period which is previously decided.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a signal processing unit and a signal processing circuit using the signal processing unit.
In particular, the present invention relates to a signal processing unit using a neural network and a signal processing circuit using the signal processing unit.

[0002]

2. Description of the Related Art In conventional information processing technology, a program created by designing an algorithm so that a target process is performed on input information is incorporated in a device in advance, and the input information is stored in the device. High-speed and large-volume processing is performed according to an incorporated algorithm.

In such an information processing technique, an algorithm is designed in advance by assuming the structure of the information to be processed. Therefore, it is possible to reliably perform processing on the assumed information. If the structure is completely unexpected, it cannot be processed.

[0004] On the other hand, in the natural world, living organisms perform information processing irrespective of such restriction conditions. A striking feature of living organisms' information processing is their ability to learn the appropriate information processing style when they encounter a new environment.

[0005] That is, information processing of living things is performed by the nervous system centered on the brain, but in living things, algorithms for information processing are not pre-installed,
It is acquired automatically through "learning." Generally, a process in which the nervous system automatically acquires an algorithm through learning is called “self-organization”.

In recent years, the neural network information processing technology, which has attracted attention, follows the self-organization mechanism in the cerebral nervous system and exceeds the limitation of "providing an algorithm in advance" which is associated with the conventional information processing technology. The aim is to build a new information processing technology.

In general, the nervous system centered on the brain is a unit of nerve cells called neurons, and a large number of these neurons are connected in a complex manner to form a neural network. Each neuron consists of a collection of branched fibers called dendrites and fibers with long side branches called axons.

[0008] A pulse-like potential change called an impulse is transferred from another neuron to the dendrite, and the impulse propagates from the dendrite to the axon. When the impulse propagates through the axon and reaches the synapse, which is the output end of the axon, it releases neurotransmitters such as glutamate and acetylcholine to dendrites and cells of other neurons that connect to the synapse.

Synapses have the property of changing their transmission efficiency and changing the level of impulses transmitted to other neurons in accordance with the history of signal transmission that has occurred there (synaptic plasticity). Therefore, an impulse having a size corresponding to the history of signal transmission is delivered to another neuron that connects to the synapse.

Information processing in the nervous system centered on the brain is performed by accumulation of such signal transmission from neuron to neuron over the entire network. The plastic change in transmission efficiency at each synapse is considered to be an elementary process of self-organization of the entire nervous system.

When a neural network is constructed by applying such information processing in the nervous system centered on the brain, the neural network is configured as having a plurality of signal transmission units corresponding to each neuron. However, in general, simply changing the transmission efficiency at each signal transmission site corresponding to a plurality of synapses in accordance with the signal passing through itself does not have any purpose to change the signal flow as a whole neural network. .

In other words, in order for the change in the transmission efficiency of the signal transmission portion of each signal transmission unit to correspond to the self-organization, the change in the transmission efficiency of each signal transmission portion in each unit may occur differently. Instead, each signal transmission part is performed depending on the change state of the transmission efficiency of the other signal transmission part while referring to the change of the transmission efficiency of the other signal transmission part configured on the same unit. is necessary.

[0013] In the nervous system of actual organisms, it has not yet been elucidated how the individual synapses cross-reference and relate to the self-organization of the nervous system. However, between the synapses formed on the same neuron, Winner-Take-All (hereinafter, referred to as
Called WTA. ) Type of competitive interaction, that is, when a specific pattern is input, a competitive phenomenon occurs between synapses, and only the specific synapses are selectively strengthened, while the other synapses are suppressed. It is known to be useful for organizing (S. Tanaka. Neural Networ
ks. Vol. 3, P625 (1990)).

In response to this, in neural network information processing technology, it is useful for realizing self-organization to incorporate a WTA-type competitive action into a signal transmission site corresponding to a synapse of each signal transmission unit. (S. Grossberg. Biological Cbernetic
s. Vol.23, P121 (1976)).

Hereinafter, WTA-type competition will be described by taking the role and mechanism of WTA-type competition in the nervous system of an organism as an example. Here, the neuron on the sender side of the signal via the synapse is called a presynaptic neuron, and the neuron on the receiver side is called a post-synaptic neuron.

Since each neuron receives signal transmission from a plurality of neurons, when one of the neurons constituting the circuit network is noted as a post-synaptic neuron, its dendrites have pre-synaptic neurons. Multiple synapses are formed to receive signals from neurons.

Here, if the presynaptic neurons that transmit signals to the post-synaptic neurons of interest code each of some events, if the signals are transmitted from these pre-synaptic neurons to the post-synaptic neurons, Of the transmitted signals, the synapse of the presynaptic neuron that transmitted the strongest signal is selected by WTA-like competition, so only the transmission efficiency of the synapse of the presynaptic neuron that transmitted the strongest signal is enhanced. This means that post-synaptic neurons selectively code the information carried by this strongest signal. These properties explain the mechanism by which feature-extracting cells are self-organized in the nervous system of an organism.

Recently, it has been theoretically suggested that WTA-like competition between synapses can occur through biochemical processes in postsynaptic neurons (H. Okamoto. And K. et al.
Ichikawa. Physical ReView E. Vol.49, P3412
(1994)). The following briefly describes the biochemical processes in postsynaptic neurons.

The synapse has two types of enzymes, a first enzyme and a second enzyme, for achieving WTA-like competition between synapses. The first enzyme is soluble in the non-activated state, but has a property of being membrane-bound in the activated state, and has an autocatalytic action of activating itself by acting as a catalyst. The second enzyme has a property of catalyzing the deactivation of the first enzyme.

At an initial time, when a signal is transmitted from each of a plurality of presynaptic neurons connected to the postsynaptic neuron to each synapse of the postsynaptic neuron,
According to the strength of the transmitted signal, the first enzyme in the inactive state, which is suspended in the cytoplasm by diffusion and transport, is activated near the synapse that has emitted the signal, and emits the signal. Binds to synapses.

Since the first enzyme has an autocatalytic action,
By autocatalysis of the first enzyme bound to the synapse, the inactive first enzyme floating between the synapses is activated to change into membrane-bound and try to take it into itself.

As a result of the incorporation of the inactive first enzyme by each synapse, competition between the synapses over the inactivated first enzyme occurs.
Begins. After a sufficient time has elapsed, the activated first enzyme concentrates at the synapse where the activation of the first enzyme was most predominant at the initial time.

That is, at this initial time, the synapse in which the activation of the first enzyme is most predominant is the sole winner, and finally the first synapse in the activated state is activated.
Will take away all of the enzyme.

In the WTA process in the nervous system of an organism, an output signal output () at an end time with respect to an input signal input (an intensity of a signal transmitted at a plurality of synapses for receiving a signal from a presynaptic neuron) at an initial time. It is thought that not only the correspondence of which synapse is selected) but also the middle is important.

That is, it is considered that the process itself of competition between synapses for the inactive first enzyme also plays an important role in the self-organization of the nervous system. Therefore, in engineeringly reproducing the self-organization, W
It is necessary to reproduce the TA competition process as closely as possible in vivo.

[0026]

However, in the conventional neural network information processing technology, WT
In incorporating the process A, only the correspondence between the input signal input at the initial time and the output signal output at the end time is incorporated into the learning rule, and the WTA competition process occurring in the living body is subjected to a chemical reaction similar to ecology. There is no device realized using the process.

According to the present invention, there is provided a signal processing apparatus capable of electrochemically implementing a WTA competition process utilizing autocatalysis and diffusion transport, such as a WTA competition process occurring between synapses in a living body, and performing signal processing. It is an object to provide a signal processing circuit using a unit and a signal processing unit.

[0028]

According to a first aspect of the present invention, there is provided a signal processing unit comprising: a charge-dispersing medium in which at least one of a positively charged body and a negatively charged body is dispersed; A pair of electrodes provided and at least one of the pair of electrodes are provided, and when a voltage is applied in a direction to capture the charged body, the charged body in the charge dispersing medium is captured and the conductivity thereof is reduced. A charge capturing layer that raises the charge and returns the captured charge to the charge dispersion medium when a voltage is applied to the pair of electrodes in a reverse direction.

The pair of electrodes are provided in the charge dispersion medium, and form an electric field in the charge dispersion medium by applying a voltage. The charge dispersing medium has at least one of a positively charged member and a negatively charged member dispersed therein, and the charged member freely moves in the charge dispersing medium.

Examples of such a charge dispersing medium include an anion solution in which a negatively charged charged body such as negatively charged particles and negative ions is dispersed, and a positively charged particle and positive ions such as positive ions. Examples include a cation solution in which a charged body charged to a charge is dispersed, and an electrolytic solution containing both a charged body charged to a negative charge and a charged body charged to a positive charge.

At least one of the pair of electrodes is provided with a charged body capturing layer for capturing a charged body in the charge dispersion medium. When a voltage is applied to the pair of electrodes in the charge dispersing medium, an electric field is formed, so that the charged body in the charge dispersing medium is attracted to the electrode side of the charge opposite to its own charge against the potential gradient.

The charged body capturing layer is provided on the electrode side to which the charged body is attracted. The charged body attracted to the electrode side is captured inside, and the conductance of the charged body capturing layer itself is determined by the amount of the captured charged body. Just raise. Therefore, the capability of capturing the charged body is higher than that before capturing the charged body by the amount of capturing the charged body.

When a reverse voltage is applied to the electrode, the charged body is attracted to the other electrode side, so that the charged body capturing layer returns the captured charged body to the charge dispersion medium. Therefore, when a reverse voltage is applied for a sufficient time, all the charged bodies captured by the charged body capturing layer are finally returned to the charged body dispersion medium, and the conductance before the charged body is captured is obtained. As a result, the charged body returned from the charged body trapping layer into the charged body dispersion medium can freely move again in the charge dispersion medium.

For example, when the charge dispersion medium is an anion solution in which anions are dispersed, the charged body capturing layer captures anions attracted to the positive electrode by applying a voltage.
As a result, the conductance of the charged body trapping layer itself increases, so that the anion trapping ability is higher than before the anion trapping. Conversely, the same applies when the charge dispersion medium is a cation solution in which cations are dispersed.

As such a charged body trapping layer, an anion-trapping type conductive polymer composed of a polymer that takes on an anion by applying a voltage and becomes conductive (hereinafter referred to as an anion-trapping type conductive polymer) is used. A conductive polymer film, a cation-capturing conductive polymer film composed of a polymer that captures cations by applying voltage and becomes conductive (hereinafter, referred to as a cation-capturing conductive polymer), and It is composed of a both-supplementary conductive polymer film composed of a polymer that captures both anions and cations by applying a voltage and has conductivity (hereinafter referred to as a both-supplementary conductive polymer). be able to.

If the charged body dispersed in the charge dispersing medium is negatively charged, for example, negatively charged particles or anions, the anion-trapping type conductive polymer film is used as a positively charged body trapping layer. Provided on the electrode side, conversely, if the charged body dispersed in the charge dispersion medium is positively charged such as positively charged particles or cations, the cation-trapping type conductive polymer film is charged It is provided on the negative electrode side as a supplementary layer.

When the charge dispersion medium is a medium containing both a negatively charged charged body and a positively charged charged body such as an electrolytic solution, a conductive material is applied to one of the pair of electrodes. A polymer membrane (anion-trapping type, cation-trapping type, or both-trapping type) is provided, or an anion-trapping type conductive polymer film is provided on one of a pair of electrodes and a cation-trapping type conductive polymer is formed on the other. It is also possible to provide a film, or to provide both complementary type conductive polymer films on both of the pair of electrodes so as to capture the charged body at both electrodes.

With such a configuration, if the amount of the charged body captured by the charged body capturing layer increases, the more
A signal processing unit capable of performing signal processing similar to signal processing at a synapse in a living body in which the ability to capture a charged body is increased is obtained. Furthermore, the charged body capturing layer also has the property of returning the charged body captured by applying a reverse voltage to the charge dispersion medium, so that the charged body can be reversibly captured by switching the voltage applied to the signal processing unit. Also in this respect, signal processing similar to signal processing at synapses in a living body can be performed.

As described above, in the first aspect, the self-catalytic action of the charged body, which causes the charged body itself to attract another charged body, and the movement (ie, diffusion) of the charged body in the charge dispersion medium. (Transport) and the reversible charged body capturing action of the charged body capturing layer can form a signal processing unit that performs signal processing similar to signal processing at a synapse in a living body.

The signal processing unit having the structure described in claim 1 is a signal processing unit which is considered to occur at a synapse in a living body.
It can be used as a unit of a signal processing circuit that realizes a TA competition process.

The reason will be described below. The voltage applied to the charged body capturing layer of the signal processing unit is V ₁ , the voltage applied to the charge dispersion medium is V ₂ , the conductance of the charged body capturing layer is G ₁ , and the conductance of the electrolyte solution is G ₂ . Since the voltage applied to the entire signal processing unit is distributed to V ₁ and V ₂ , V ₁ : V ₂ = 1 / G ₁ : 1 / G ₂ holds.

Assuming that the amount of charge in the charge dispersing medium is much larger than the amount that can be captured by the charged body capturing layer, the conductance G ₂ of the electrolyte solution is always constant, that is, G ₂ =
Const. It is. V ₁ : V ₂ = 1 / G ₁ : 1 / G ₂ , regardless of whether the voltage applied to the entire unit is constant or not, the application time of the voltage applied to the charged body capturing layer since the charging member capturing layer the longer captures correspondingly many charged body conductance G ₁ rises, the voltage V ₂ applied to the charge dispersion medium is increased. That is, it can be said that the rate of increase in the charged body capturing ability is proportional to the voltage V ₂ applied to the charge dispersion medium. Therefore, it is understood that the capture of the charged body occurs autocatalytically.

For the above reasons, the configuration of the signal processing unit according to the first aspect of the present invention is used as a unit of a signal processing circuit for realizing a WTA competition process in order to introduce self-catalytic property which is a basic element for realizing the WTA process. Available.

In a signal processing circuit according to a second aspect of the present invention, a plurality of the signal processing units according to the first aspect are provided in the same charge dispersion medium, and each of the signal processing units is provided based on information input from the outside. A first power supply for applying a voltage of the determined magnitude to the direction in which the charged body capturing layer captures the charged body, and a voltage of a predetermined constant magnitude for all of the plurality of signal processing units. A second application is performed in a direction in which the charged body is returned from the charged body capturing layer into the charge dispersion medium.
And the first power supply so that at least a part of the charged bodies in the charged body capturing layer that has captured the most charged bodies by applying the voltage from the first power supply remains in the charged body capturing layer. A switch mechanism for switching between a power supply and the second power supply at predetermined time intervals.

The first power source applies a voltage of a magnitude determined for each signal processing unit to the electrode provided with the charged body capturing layer in a direction in which the charged body capturing layer captures the charged body. Assuming that the voltage applied by the first power supply is an input voltage,
Since different input voltages are applied to the respective signal processing units by the first power supply, the ability of the charged body capturing layer provided on each electrode to capture the charged bodies also increases in the descending order of the voltage value.

The second power supply applies a constant voltage in a direction opposite to that of the first power supply to each signal processing unit. As a result, the charged body is partially returned to the charge dispersion medium from the charged body trapping layer provided in each unit.

The switch mechanism switches the above-mentioned two types of power sources at predetermined time intervals. At least when a charged body is returned from each charged body capturing layer to the charge dispersion medium by application of the second power supply, To the extent that not all charged bodies in the charged body capturing layer that captured the most charged bodies are released from the charged body capturing layer, in other words, of the charged bodies in the charged body capturing layer that captured the most charged bodies Is switched between the first power supply and the second power supply so that at least a part of the first power supply remains in the charged body capturing layer.

When the signal processing circuit having such a configuration is operated, in the initial stage, the charged body trapping layer is applied to the charge dispersing medium with a strength corresponding to the magnitude of the input voltage applied by the first power supply. , The amount of the charged body corresponding to the magnitude of the applied voltage is captured by the charged body capturing layer of each signal processing unit.

When the power supply is switched to the second power supply by the switch mechanism after the lapse of a predetermined time, a constant reverse voltage is applied to each signal processing unit for a predetermined time.
As a result, the same amount of the charged body captured by the charged body capturing layer of each signal processing unit is returned to the charge dispersion medium from the charged body capturing layer of each signal processing unit. At this time, preferably, the charged body remains in the charged body capturing layer in the signal processing unit to which a large input voltage is applied by the first power source, but the charged body is captured in the signal processing unit to which a small input voltage is applied. The magnitude and time of the reverse voltage are adjusted so that all the charged bodies in the layer return to the charge dispersion medium.

When the power supply is again switched to the first power supply by the switch mechanism, of the charged body capturing layers of each signal processing unit, the charged body capturing layer in which the charged body remains inside has higher conductance than the initial stage. Therefore, the ability to capture a charged body is stronger than in the initial stage, and captures more charged bodies than in the initial stage.

That is, in the signal processing unit having a high input voltage, the more the switching of the power supply by the switch mechanism, the more the charged body remaining in the charged body capturing layer and the stronger the capturing ability of the charged body. Since the signal processing unit having a low input voltage has no charged body remaining in the charged body capturing layer, the ability of capturing the charged body does not change.

Further, the more the power supply is switched by the switch mechanism, the more the charged body is captured by the signal processing unit having a higher input voltage, so that the concentration of the charged body in the charge dispersion medium gradually decreases. Therefore, when the voltage is applied by the second power supply, the charged body returned to the charge dispersion medium by the signal processing unit to which the low input voltage is applied is also dispersed by diffusion, so that the signal processing unit having the gradually increased input voltage , And eventually all the charged bodies are captured by the signal processing unit having the largest input voltage.

As described above, by switching between the first power supply and the second power supply by the switch mechanism, the charged body moving in the charged body dispersion medium apparently charges the charged body provided in each signal processing unit. Electrochemically achieving a WTA-like operation in which the body-trapping layers compete with each other and compete for each other, and finally the signal processing unit to which the strongest voltage is applied by the first power supply captures all charged bodies. Can be.

[0054]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS. FIG. 1 is an explanatory diagram schematically showing a signal processing device in which the signal processing circuit of the present invention is implemented. In this signal processing device, a plurality of signal processing units 10 are configured as one device using the same charge dispersion medium, with the signal processing unit 10 as one unit as shown in FIG.

Note that the signal processing apparatus of the present embodiment
Power supply provided for each of the number of signal processing units 10
The unit 12 is configured as one power supply device 20. Ma
Anion solution 14 in which anions are dispersed as a charged body;
Is used as a charge dispersion medium. This anion solution 14
Is CH, an organic solvent_ThreeElectrolyte in CN (C_TwoH
_Five)_FourNCLO_FourAnd ClO as an anion_Four ^-I
ON is generated.

The signal processing unit 10 shown in FIG. 2 has a first electrode 18a having a polypyrrole layer 16 which is a kind of conductive polymer film formed on the surface thereof as a charged body capturing layer, and a polypyrrole layer 16 formed on the surface. Not the second electrode 18b
And a power supply unit 12 for applying a voltage to the pair of electrodes. In this signal processing unit 10, the polypyrrole layer 16 of the first electrode 18a corresponds to a synapse in the biological nervous system, and the ClO ₄ ⁻ ions in the anion solution 14 realize the WTA-like competition between synapses as described above. To the first enzyme to perform.

The anion solution 14 is filled between the pair of electrodes 18a and 18b, and the power supply unit 12 includes: a first power supply 15a for applying a voltage so that the first electrode 18a has a positive polarity; Conversely, a second power supply 15b for applying a reverse voltage so that the first electrode 18a becomes a negative electrode, and a first power supply 15
a and a switch mechanism 17 for switching the connection between the second power supply 15a and the second power supply 15b.

The switch mechanism 17 switches the first power supply 15 every time a predetermined time elapses by the voltage control unit 22 shown in FIG.
a and switches the connection between the second power supply 15b and the second power supply 15b. In the present embodiment, the period is 50 Hz (that is, 0.0 Hz).
The connection between the first power supply 15a and the second power supply 15b is switched every second).

[0059] In the signal processing unit 10 having such a configuration, when the first power source 15a to a pair of electrodes are connected, since the first electrode 18a becomes positive, ClO ₄ in anion solution 14 ^- ions move toward the first electrode 18a. Since an electric field is formed on the first electrode 18a side, ClO ₄ ⁻ ions that have moved toward the first electrode 18a enter the polypyrrole layer 16 and are trapped (that is, doped). Becomes The ClO ₄ ^- by doping ion conductance of polypyrrole layer 16 also increases ClO ₄ ^- is ability to capture ions further improved.

When the second power supply 15b is connected to the pair of electrodes, the second electrode 18b serves as a positive electrode, so that ClO ₄ ^- ions in the anion solution 14 Move towards. The ClO ₄ ⁻ ions doped into the polypyrrole layer 16 on the first electrode 18a side are
The anion solution 14 is gradually separated from the polypyrrole layer 16.
Disperse inside. In this embodiment, since the second electrode 18b does not include a layer such as a polypyrrole layer that captures ClO ₄ ⁻ ions, the conductance of the second electrode 18b does not change.

Therefore, when the second power supply 15b is connected, the conductance of the polypyrrole layer 16 which is increased by taking in ClO ₄ ^- ions into the polypyrrole layer 16 gradually decreases. All ClO ₄ ^- ions in 16 are anion solutions 1
4, the conductance of the polypyrrole layer 16 returns to its own conductance.

The signal processing device of the present embodiment includes a plurality of signal processing units 10 operating as described above and a voltage control unit 22. For example, the voltage control unit 22 outputs a plurality of input data having different weights, such as a plurality of input data weighted according to a certain rule such as image data and a plurality of input data weighted irregularly. On the other hand, the signal is converted into a voltage value having a magnitude corresponding to each weight and output to the power supply device 20 of the signal processing unit 10.

The power supply device 20 according to the present embodiment is provided for each signal processing unit 10, and each signal processing unit 10 is arranged such that the first electrode 18 a of each signal processing unit 10 has a positive polarity. A first power supply unit (not shown) composed of a plurality of first power supplies 15a (see FIG. 2) for applying a voltage corresponding to the first and second power supplies 15a (see FIG. 2). A second power supply unit (not shown) for applying a uniform reverse voltage to all the signal processing units 10 so that the first power supply unit and the second power supply unit have a frequency of 50 Hz.
, Ie, a switch mechanism 17 (see FIG. 2) that switches every 0.01 seconds.

In the signal processing device having such a configuration, when data including a plurality of data of different sizes, such as image data, is input, the voltage control unit 22 reduces the voltage value to a voltage value corresponding to each data. It is converted and output to the power supply device 20.

In the power supply device 20, the plurality of input voltage values correspond to the respective output voltages of the plurality of first power supplies 15a constituting the first power supply section. When the connection to the first power supply unit is switched by the switch mechanism 17 in the power supply unit 20, the power supply unit 20 is turned on by the first power supply 15a provided for each signal processing unit 10.
To each of the first electrodes 18a, a voltage of a magnitude individually determined based on input data is applied. When the connection is switched to the second power supply unit, the power supply device 20
Applies a uniform reverse voltage to each signal processing unit 10.

The switch mechanism 17 switches between the first power supply unit and the second power supply unit a plurality of times to change the direction in which the voltage is applied.
Since the conductance of all the electrodes 18a is the same, an electric field having a strength corresponding to the magnitude of the applied voltage is formed on each first electrode 18a.
In the polypyrrole layer 16 of the electrode 18a, ClO ₄ ⁻ ions are doped in an amount corresponding to the weight of the input data.

When the connection is switched to the second power supply section by the switch mechanism 17 and a reverse voltage is applied, the ClO ₄ ^- ions in the polypyrrole layer 16 of each first electrode 18a are gradually removed from the polypyrrole layer 16. I will leave.

At this time, the magnitude of the reverse voltage applied by the second power supply unit is such that ClO ₄ ⁻ ions in the polypyrrole layer 16 of the first electrode 18a to which at least the highest voltage is applied by the first power supply unit are reduced. The remaining size.

When the connection is switched again to the first power supply unit by the switch mechanism 17, the polypyrrole layer 16 in which the ClO ₄ ^- ions remain inside of the polypyrrole layer 16 of each signal processing unit 10 is in the initial stage. Because of the higher conductance, the ability to capture ClO ₄ ^- ions is stronger than in the earlier stages, thus doping more ClO ₄ ^- ions than in the earlier stages.

That is, in the signal processing unit 10 having a high input voltage, the more the power supply is switched by the switch mechanism 17, the more ClO ₄ ⁻ remaining in the polypyrrole layer 16.
As the amount of ions increases, the ability to capture ClO ₄ ^- ions increases.

[0071] Further, ClO to higher input voltages the signal processing unit 10 performs be performed switching of power by the switch mechanism 17 ₄ ^- anion solution 1 because the ions are trapped
The concentration of ClO ₄ ^- ions in ₄ gradually decreases. Therefore, when the connection is switched to the second power supply unit, the ClO ₄ ⁻ ions returned to the anion solution 14 by the signal processing unit 10 to which a low input voltage is applied are also gradually transferred to the signal processing unit 10 having a high input voltage. All the ClO ₄ ⁻ ions are finally captured by the signal processing unit 10 to which the highest voltage is input by the first power supply unit.

As described above, the first mechanism by the switch mechanism 17
By switching between the power supply unit and the second power supply unit, apparently, the polypyrrole layer 16 provided in each signal processing unit 10 is provided.
Mutually compete for ClO ₄ ⁻ ions moving in the anion solution 14 and compete for each other. Eventually, the signal processing unit 10 to which the strongest voltage is applied by the first power supply unit causes all the Cl
A WTA-like operation of capturing O ₄ ^- ions is electrochemically realized.

Here, the result of analyzing the operation of the signal processing device having the above configuration by computer simulation will be described below.

First, in order to construct a mathematical model of the signal processing device of the present embodiment, the anion trapping rate of the polypyrrole layer used as the charged layer trapping layer and the conductance of the polypyrrole layer used in the signal processing device of FIG. The relationship was considered.

FIG. 3 shows the relationship between the anion trapping rate of the polypyrrole layer and the conductance of the polypyrrole layer, which is shown by BJ Feldman, P. Burgmayer, ann.
d RWMurray, J. Am. Chem. Soc. 1985, 107 , P872-878
Are described. In FIG. 3, the vertical axis is the conductance G (S / m) of the polypyrrole layer, and the horizontal axis is the anion trapping rate X per polymer in the polypyrrole layer.

As shown in FIG. 3, it can be said that the relationship between the conductance G of the polypyrrole layer and the anion trapping rate X is almost proportional in the low trapping rate region (X = 0 to about 0.16). Anion trapping rate X of this polypyrrole layer 16
Is based on the electric action, so the above proportional relationship can be applied not only to polypyrrole but also to general conductive polymer materials.

Therefore, the relationship between the conductance G ₁ of the conductive polymer film and the anion trapping rate X can be expressed by the following equation (1).

G ₁ = kX (1) From FIG. 3, the value of k was calculated as k = 93 (s / m).

Since the anion trapping rate X represents the number of anions adsorbed to one unit of the polymer constituting the conductive polymer film, when A ₁ is the volume of the conductive polymer film, The anion trapping rate X can be represented by the following equation (2).

[0080]

(Equation 1)

Here, Q ₁ is the charge per unit volume of the anion trapped in the conductive polymer film, n is the valence of the anion, e is the elementary charge, ω is the mass of the conductive polymer film, _A is Avogadro's number and M is molecular weight. From the equations (1) and (2), the conductance G ₁ of the conductive polymer film is obtained.
Can be expressed as in equation (3).

[0082]

(Equation 2)

Since (kA ₁ · M / neωN _A ) is a constant, if (kA ₁ · M / neωN _A ) = k ₁ , the above equation (3) becomes the following equation (3) ′. Can be represented.

G ₁ = k ₁ · Q ₁ (3) ′ Here, assuming that the amount of the electrolyte in the anion solution 14 is excessive compared to that of the polymer, the conductance G ₂ of the electrolyte solution is always constant. It can be said that there is. The conductance of the first power supply 15a is represented by G ₍₊₎ , and the first power supply 15a
When but the voltage supplied to the V _(+), when it is switched to the first power supply unit by the switch mechanism 17, the first power source 15a provided corresponding to each signal processing unit 10, the signal processing unit The voltage V ₁ applied to the conductive polymer film of the first electrode of No. 10 can be represented by the following equation (4), and the voltage V ₂ applied to the anion solution 14 is It can be expressed as the equation (5).

V ₁ = G ₍₊₎ G ₂ V ₍₊₎ / ((G ₍₊₎ + G ₂ ) G ₁ + G ₍₊₎ G ₂ ) (4) V ₂ = G ₍₊₎ G _{1 V (+) / ((G} (+) + G 2) G 1 + G (+) G 2) also ... (5), the area of the adjacent surfaces of adjacent units S, the anion in the anion solution 14 speed since the V _3, when the charge amount per unit volume of the anion in the anion solution 14 was Q _2, is _{a 1 ΔQ 1 = V 3 ΔtSQ} 2,
When being switched to the first power supply unit by the switch mechanism 17, the charge amount ΔQ _{1 of the} anion trapped in the conductive polymer film during the short time Δt is expressed by the following equation (6). Can be represented.

ΔQ ₁ / Δt = (V ₃ SQ ₂ ) / A ₁ (6) The velocity of the anion in the solution is V ₃ , the mobility of the anion in the anion solution 14 is μ ₂ , the first power supply when the electric field strength when switched to part and E _2, V ₃ = a μ ₂ E _2, E ₂ is E ₂ = V ₂ / I ₂ when the thickness of the anion solution 14 and I ₂ In addition, when the thickness of the conductive polymer film is I ₁ , A ₁ = SI ₁ , so that the above equation (6) can be expressed as the following equation (7).

[0087] By substituting the equations (3) and (5) into the equation (7),
The following equation (8) is obtained.

[0088]

(Equation 3)

On the other hand, when the switch mechanism 17 switches to the second power supply unit, the charge amount ΔQ ₃ of the anion in the conductive polymer film released during the short time Δt is determined by the second power supply unit. By setting the conductance to G _(-) , the voltage supplied by the second power supply unit to V _(-) , and the mobility of the anion in the conductive polymer film to μ ₁ , the following calculation is performed. Equation (9) is obtained.

[0090]

(Equation 4)

Next, parameters and variables in the i-th signal processing unit i (where i is a positive integer) among the plurality of signal processing units 10 constituting the signal processing device will be described. The parameters and variables in the signal processing unit i are expressed as a function of i, for example, Q ₁ (i).

First, the flow density J of anions due to the concentration gradient from the signal processing unit j to the signal processing unit i
_{(i, j)} represents the diffusion coefficient of anions in the anion solution 14 as D (that is, Einstein's relation D = μk
_B T, where T is absolute temperature and k _B is any 2 for simplicity
The set of Boltzmann coefficients. ) When the volume of the anion solution 14 for one unit is A ₂ and the distance between the signal processing units 10 is always d, it can be expressed by the following equation (11).

J _{(i, j)} = DS (Q _{2 (j)} −Q _{2 (i)} ) / A ₂ d (11) Next, when the first power supply unit is switched, the i-th position The rate of change dQ _{1 (i)} / dt of the amount of charge Q _{1 (i)} per unit volume of anions trapped in the conductive polymer membrane of the signal processing unit i to be processed, and the unit volume of anions in the anion solution 14 per unit volume change rate dQ 2 of the charge amount _{Q 2 (i) (i)} / Request and dt.

First, from the equation (8), the amount of charge per unit volume of the anion trapped in the conductive polymer film of the i-th signal processing unit i when switching to the first power supply unit is obtained. The rate of change dQ _{1 (i)} / dt of Q _{1 (i)} can be expressed by the following equation (12).

[0095]

(Equation 5)

Further, according to the equations (8) and (11), when the first power supply unit is switched, the unit volume of anions in the anion solution 14 in the anion solution 14 of the i-th signal processing unit i is determined. The rate of change of the charge amount Q _{2 (i)} dQ _{2 (i)} /
dt can be expressed as in the following equation (13).

[0097]

(Equation 6)

Further, when switching to the second power supply unit is performed.
In addition, the high conductivity of the i-th signal processing unit i
Charge Q per unit volume of anions in the membrane_{3 (i)}Strange
Conversion rate dQ_{3 (i)}/ Dt and the anion in the anion solution 14
Charge Q per unit volume of _{4 (i)}Rate of change dQ_{4 (i)}/ D
and t.

While the first power supply unit applies a voltage of a magnitude corresponding to each unit, the second power supply unit applies a constant voltage to all the signal processing units 10, so that V
_{(-) ( i)} = V _(-) . From the equation (8), the change in the charge amount Q _{3 (i)} per unit volume of the anion in the conductive polymer film of the i-th signal processing unit i when switching to the second power supply unit is made. The rate dQ _{3 (i)} / dt is given by the following (1)
4) It can be expressed as in the following equation.

[0100]

(Equation 7)

Further, according to the formulas (8) and (11), when switching to the second power supply unit is performed, the unit per unit volume of anions in the anion solution 14 of the i-th signal processing unit i is determined. The rate of change of the charge amount Q _{4 (i)} dQ _{4 (i)} /
dt can be expressed as in the following equation (15).

[0102]

(Equation 8)

Since the switch mechanism 17 switches between the first power supply unit and the second power supply unit at a cycle of 50 Hz, the direction of the voltage is switched every 0.01 seconds. Furthermore, the total charge C of the anion in each signal processing unit 10 at the initial time is equal to the charge (A ₁ Q _{1 (i)} ) of the anion in the conductive polymer film at the initial time (t = 0). And the sum of the charge amount of the anion in the anion solution 14 corresponding to one unit (A ₂ Q _{2 (i)} ), that is, the following equation (16).

A ₁ Q _{1 (i)} + A ₂ Q _{2 (i)} = C (16) Each signal processing is performed by calculating the differential equations (12) to (15) by a computer and numerically solving them. It was investigated what kind of interaction would occur between the signal processing units 10 by applying a voltage of a magnitude corresponding to each unit 10. Table 1 shows the values of the parameters used.

[0105]

[Table 1]

FIG. 4 shows an example in which the total number of units N is 5
Graphed calculation results of change in charge amount to Q ₁ per unit volume of anion trapped in the conductive polymer film of each signal processing unit 10 (I) ~ every (V) where it is shown. In addition,
Values are approximations. In FIG. 4, the vertical axis represents the amount of charge Q ₁ (C / m) per unit volume of the anion captured by the conductive polymer film, and the horizontal axis represents time t (s).

FIG. 4 shows that the charge per unit volume of anions trapped in the conductive polymer film of the signal processing unit 10 to which the largest voltage was applied by the first electrode unit among the five signal processing units 10. Q ₁ (V) increases over time.

The charge Q ₁ (I) per unit volume of anions trapped in the conductive polymer film of the signal processing unit 10 other than the signal processing unit 10 to which the largest voltage is applied.
Some of (IV) slightly increase at the beginning, but decrease as time elapses, and become zero after about 10 seconds. Thus, the signal processing units 1 other than the signal processing unit 10 to which the largest voltage is applied
It can be seen that all the anions captured by the conductive polymer film of No. 0 have been removed.

That is, from this result, it is clear that WTA-like competition of anions occurs between the units based on the difference in the magnitude of the applied voltage between the first electrode units. In the above, the case where the total number of units N is 5 has been described.
Similar results are obtained when the total number of units N is changed.

Therefore, according to the signal processing device of the present embodiment, the signal processing unit 10 to which the highest voltage is applied by the first power supply unit among the plurality of signal processing units 10 is selected, and finally selected. Signal processing unit 10
Only in the case of above, trapping of anions in the conductive polymer film is achieved.

As described above, the signal processing device of the present embodiment can realize a WTA-like competition phenomenon of anions regardless of the total number of units, so that a WTA process useful for constructing a neural network information processing technology can be realized. realizable.

In the above embodiment, (C ₂ H ₅ ) ₄ NC which is an electrolyte in CH ₃ CN is used as a charge dispersion medium.
Although the anion solution 14 in which 10 ₄ was dispersed and ClO ₄ ⁻ ions were generated as anions was used, the present invention is not limited to this, and another anion solution may be used. Of course, the anion in the anion solution is not limited to a monovalent anion, but may be a divalent or higher valent anion.

In the signal processing device of the above-described embodiment, an anion solution in which anions are dispersed is used as a charge dispersing medium as a charge dispersing medium. It is also possible to use.

Although polypyrrole, which is a kind of conductive polymer film, was used as the charged body capturing layer, it is not limited to this. Others that can be used as the charged body capturing layer include, for example, polyacetylene-based, polydiacetylene-based, polyheptadiene-based, polypyrrole-based, polythiophene-based, polyaniline-based, polyphenylenevinylene-based, polythiophenylenevinylene-based, and polyisothianephthene. System, polyisonaphthothiophene, polyparaphenylene, polyphenylene sulfide, polyphenylene oxide, polyfuran, polyphenanthrene, polyselenophene, polytellurophen, polyazulene, polyindene, polyindole, poly Phthalocyanine type, polyacene type, polyacenoacene type, ponaphthylene type, polyanthracene type, polyperinaphthalene type, polybiphenylene type, polypyridinopyridine type, polycyanogen type, polyarene There are various types of one-dimensional conductive polymers such as a non-conductive type, ladder polymers, pyropolymers, and precursors of two-dimensional conductive polymers such as graphite (ie, monomers and prepolymers). .

In the above one-dimensional conductive polymer,
The polypyrrole-based one-dimensional conductive polymer and the polyaniline-based one-dimensional conductive polymer are cation-trapping types, the polyviridinopyridine-based ones are cation-trapping types, and the polythiophene-based ones trap both cations and anions. Since both are trapping types, the type and concentration of the charge dispersing medium,
It is advisable to adjust various conditions such as a power supply switching timing.

Further, as another embodiment, it is possible to provide a structure in which a charged body capturing layer is provided on both the first electrode and the second electrode. In this case, it is preferable to use a solution in which a positively charged body and a negatively charged body coexist, such as an electrolyte solution, as the charge dispersion medium.

[0117]

As described above, according to the present invention, there is an effect that a signal processing unit capable of performing signal processing similar to signal processing at a synapse in a living body can be obtained.

Further, according to the present invention, apparently, the charged body trapping layers provided in the respective signal processing units compete with each other for the charged body moving in the charged body dispersion medium, and finally compete with each other. There is an effect that a signal processing device capable of electrochemically realizing a WTA-like operation in which a signal processing unit to which a strongest voltage is applied by a power supply captures all charged bodies is obtained.

[Brief description of the drawings]

FIG. 1A is a schematic configuration diagram of a signal processing device according to an embodiment of the present invention, and FIG. 1B is a diagram illustrating a first electrode 18;
It is a top view for explaining arrangement of a.

FIG. 2 is a schematic configuration diagram of a signal processing unit 10 included in the signal processing device of FIG.

FIG. 3 is a graph showing a relationship between an anion trapping rate X and a conductance G of a polypyrrole layer 16 used as a charged body trapping layer in the signal processing device of FIG.

FIG. 4 shows changes (I) to (V) in the charge amount Q ₁ per unit volume of anions trapped in the conductive polymer film of each signal processing unit 10 when the total number of units N is 5; It is a graph shown.

[Explanation of symbols]

 Reference Signs List 10 signal processing unit 12 power supply unit 14 anion solution 15a first power supply 15b second power supply 16 polypyrrole layer 17 switch mechanism 18a first electrode 18b second electrode 20 power supply device 22 voltage control unit

Claims (2)

[Claims] 1. A pair of electrodes provided in a charge dispersion medium in which at least one of a positively or negatively charged charged body is dispersed, and a pair of electrodes provided on at least one of the pair of electrodes. When a voltage is applied in the capturing direction, the charged body in the charge dispersion medium is captured to increase its own conductivity, and
A charge capturing layer for returning the captured charged body into the charge dispersion medium when a voltage is applied in the opposite direction to the pair of electrodes. 2. A plurality of the signal processing units according to claim 1 are provided in the same charge dispersion medium, and a voltage having a magnitude determined for each signal processing unit based on information input from the outside is provided. A first power source applied in a direction in which the charged body capturing layer captures the charged body; and a voltage having a predetermined constant magnitude for all of the plurality of signal processing units. And a second power supply applied in a direction returning from the charge dispersion medium to the charge dispersing medium; and at least a part of the charged bodies in the charged body capturing layer that has captured the most charged bodies by applying a voltage from the first power supply. A switch mechanism for switching between the first power supply and the second power supply at predetermined time intervals so that 残 る remains in the charged body trapping layer.