WO2016208817A1

WO2016208817A1 - Apparatus and method for interfacing key input

Info

Publication number: WO2016208817A1
Application number: PCT/KR2015/010610
Authority: WO
Inventors: 김태호; 이슬; 이동욱; 정대웅
Original assignee: 주식회사 노타
Priority date: 2015-06-22
Filing date: 2015-10-07
Publication date: 2016-12-29
Also published as: KR101653167B1

Abstract

A key input interface apparatus according to a first embodiment of the present invention comprises: an input unit for receiving touch points for a plurality of keys included in a key input apparatus; a distribution modeling unit for modeling distribution of a set of the touch points for each key, using a pre-defined probability model; and a configuration unit for configuring, on the basis of the modeled distribution, a recognition area recognized by an input for the key.

Description

Apparatus and method for interfacing keystrokes

The present invention relates to an apparatus and a method for interfacing a key input. More particularly, the present invention relates to a method of modeling a distribution of a point touched by a user as a probabilistic model, and to reduce the error rate by adaptively changing a recognition region of a key. An apparatus and method that can be used.

A portable terminal such as a smartphone is a means for receiving data, and includes various types of key input devices (for example, a QWERTY keyboard or a celestial player) that are implemented and displayed on the screen of the portable terminal.

Such a portable terminal may have a limited size screen according to its use, and a key input device may also be implemented on such a limited size screen, thereby causing various inconveniences. For example, since there is a limit on the area occupied by one key, a typo may occur frequently in which other letters are input other than a key intended by the user.

In order to reduce such a typo, a variety of techniques have been disclosed. For example, a method of reducing a typo in consideration of a speed of a key stroke of a user or a method of predicting a key intended by a current touch point based on a character input by a user is disclosed.

However, a technique of analyzing a user's touch point and setting the recognition region of the key different from the layout of the key has not been disclosed.

The problem to be solved of the present invention is a key input that can reduce the error rate by modeling the distribution of the set of touch points of the user as a probability model, and adaptively changing the recognition region of the key based on the modeled distribution It is to provide an interface device and method.

Another object of the present invention is to provide a key input interface device and method for analyzing a situation in which a user inputs a key and providing a different recognition area of the key according to the situation.

However, the problem to be solved of the present invention is not limited to those mentioned above, another problem to be solved is not mentioned can be clearly understood by those skilled in the art from the following description. will be.

In accordance with another aspect of the present invention, a key input interface device includes: an input unit configured to receive touch points for a plurality of keys included in a key input device; A distribution modeling unit modeling a distribution of the set of touch points for each of the keys by using a predefined probability model; And a setting unit configured to set a recognition area recognized as an input to the key, based on the modeled distribution.

The probabilistic model may be characterized in that it is unsupervised learning about information on which key the key intended by the touch point is among the plurality of keys.

In addition, the probability model may be characterized as being a parametric for estimating a parameter related to the probability model.

The distribution modeling unit may set an initial value of the parameter based on a touch point within a preset range in the layout of the key.

The distribution modeling unit may estimate the inverse covariance value based on a predetermined correction value when an inverse covariance value of the parameter is not estimated.

In addition, the probability model is a Gaussian mixture model (GMM), and the distribution modeling unit may estimate parameters of the Gaussian mixture model using expectation maximization (EM).

The setting unit may be configured to set the recognition area in a first layout corresponding to the first key or to a first distribution of a first key having a touch point input of the plurality of keys equal to or less than a preset number. The recognition area may be set on the basis of.

The setting unit may set the recognition area in consideration of whether the parameter is within a preset boundary.

In accordance with another aspect of the present invention, a key input interface device includes: an input unit configured to receive touch points for a plurality of keys included in a key input device; A distribution modeling unit modeling a distribution of the set of touch points for each of the keys by using a predefined probability model; A labeling unit for classifying and labeling hits and typos among the input touch points based on the modeled distribution; And a setting unit configured to set a recognition area recognized as an input to the key, based on the labeling result.

In addition, the labeling unit may select a hit if the likelihood of the input touch point is equal to or greater than a preset threshold.

The labeling unit may select a typo from the input touch points based on a predefined typo detection algorithm.

The setting unit may set the recognition area using a deep neural network based on the labeling result.

The key input interface device may further include a situation determination unit that determines a situation of a user of the key input device, wherein the input unit classifies an input touch point according to the situation, and the distribution modeling unit is configured according to the situation. The distribution may be modeled differently, and the setting unit may set the recognition region differently according to the situation.

The situation may be at least one of a movement of the user, a posture of the user, whether the user manipulates the key input device with both hands, or a temperature at which the key input device is operated.

According to a third embodiment of the present invention, a method of interfacing a key input of a key input device performed by a key input interface device comprises: receiving a touch point for a plurality of keys included in the key input device; Modeling a distribution of the set of touch points for each key using a predefined probability model; And setting a recognition region recognized as an input to the key, based on the modeled distribution.

The modeling may include setting an initial value of the parameter based on a touch point within a preset range in the layout of the key.

In addition, the modeling may include estimating the inverse covariance value based on a predetermined correction value when an inverse covariance value of the parameter is not estimated.

In addition, the probability model is a Gaussian mixture model (GMM), and the modeling may estimate parameters of the Gaussian mixture model using expectation maximization (EM).

The setting may include setting the recognition area in a first layout corresponding to the first key or pre-modeling the first key having a touch point input of the plurality of keys equal to or less than a preset number. And setting the recognition area based on one distribution.

In the setting, the recognition area may be set in consideration of whether the parameter is within a preset boundary.

A method of interfacing a key input of a key input device performed by a key input interface device according to a fourth embodiment of the present invention comprises: receiving touch points for a plurality of keys included in the key input device; Modeling a distribution of the set of touch points for each key using a predefined probability model; Labeling and classifying hitting and typos among the input touch points based on the modeled distribution; And setting a recognition area recognized as an input to the key, based on the labeling result.

In addition, the labeling may be selected as a hit if the likelihood of the input touch point is greater than or equal to a preset threshold.

In addition, the labeling may select a typo from the received touch point based on a predefined typo detection algorithm.

The setting may include setting the recognition region using a deep neural network based on the labeling result.

The key input interface method may further include determining a situation of a user of the key input device. The receiving of the key may include classifying the received touch point according to the situation, and the modeling may include the situation. According to the modeling and distribution of the distribution differently according to the situation may set the recognition area differently.

The method according to an embodiment of the present invention may be implemented by being included in a recording medium in which a computer program is recorded.

According to an embodiment of the present invention, the recognition region of the key may be adaptively changed and provided based on the distribution of the set of touch points of the user, and the recognition region of the key may be provided according to the situation in which the user inputs the key. Since the change can be provided, the error rate of the key input device can be reduced.

1 is a diagram illustrating a configuration of a key input interface device according to a first embodiment of the present invention.

2 is a diagram exemplarily illustrating a distribution of a touch point of a user according to the first exemplary embodiment of the present invention.

3 is a diagram exemplarily illustrating a distribution of touch points used to set initial parameters of a probability model, according to the first embodiment of the present invention.

4 is a diagram illustrating an example of modeling a distribution based on parameters of an estimated probability model, according to the first embodiment of the present invention.

FIG. 5 is a diagram illustrating an example of changing a recognition region of a key according to the first embodiment of the present invention.

FIG. 6 is a diagram exemplarily illustrating a procedure of a method for interfacing a key input according to the first embodiment of the present invention.

Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but can be implemented in various different forms, and only the embodiments make the disclosure of the present invention complete, and the general knowledge in the art to which the present invention belongs. It is provided to fully inform the person having the scope of the invention, which is defined only by the scope of the claims.

In describing the embodiments of the present invention, if it is determined that a detailed description of a known function or configuration may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. In addition, terms to be described below are terms defined in consideration of functions in the embodiments of the present invention, which may vary according to intentions or customs of users and operators. Therefore, the definition should be made based on the contents throughout the specification.

First, although not shown in the drawing, the key input interface device 100 according to the first embodiment of the present invention may be implemented by being included in a key input device that receives a key. Such a key input device may be input to a plurality of keys. The device may be inputted by a touch, for example, a smartphone, a smart pad, a PDA, a computer, a desktop, a laptop, a notebook, a workstation, or a server having a touch screen, but are not limited thereto.

In addition, the key input interface device 100 according to the first embodiment of the present invention may be implemented in an electronic device including a processor and a memory for storing instructions executed by the processor, but is not limited thereto.

Referring to FIG. 1, the key input interface device 100 according to the first embodiment of the present invention may include an input unit 110, a distribution modeling unit 130, or a setting unit 150. However, since this is exemplary, according to the exemplary embodiment, it may not include at least one or more of these components, or may further include other components not mentioned herein. In addition, these components may be located on the same physical device or on different physical devices.

The input unit 110 receives a touch point touched by a user from a key input device, and FIG. 2 is a diagram illustrating a touch point input by the input unit 110 on the touch screen as an example.

The touch point may be, for example, in the form of coordinates (x, y) on the touch screen included in the key input device.

In addition, the input unit 110 may receive all of the touch points input by the user from the key input device at once, and as will be described later, the distribution modeling unit 130 may model the distribution at once based on the entire touch points. Batch learning. However, this is only an example and the spirit of the present invention is not limited thereto. That is, according to the exemplary embodiment, the input unit 110 may receive input of some of the touch points input by the user. Accordingly, the distribution modeling unit 130 may model the distribution several times based on the touch points. It may also be online learning.

The distribution modeling unit 130 models each key using a predefined probability model based on a distribution of touch points input by the input unit 110.

In more detail, the data used for modeling by the distribution modeling unit 130 may be a set of all input touch points, that is, touch points. In addition, the distribution modeling unit may model the distribution of the touch points input for each key, and the distribution of the touch points input for each key is shown in FIG. 2.

Here, the probability model used by the distribution modeling unit 130 for modeling may be characterized as being a parametric for estimating a parameter related to the probability model. In this case, the estimated parameter may include, for example, an average or a covariance. In addition, the probabilistic model may be characterized in that the user estimates a parameter without information about a key intended by a touch point, that is, unsupervised learning.

The probability model may be, for example, a Gaussian mixture model (GMM), and the distribution modeling unit 130 estimates the parameters of the Gaussian mixture model by using an expectation maximization (EM). This can be done in more detail below.

First, the distribution modeling unit 130 sets an initial value of a parameter related to a Gaussian mixture model, and may use Equation 1 below.

Here, Θ represents the aforementioned parameter and x represents the input touch point. In addition, since α _i is a predefined weight, the right side of Equation 1 represents a sum in which weights of M Gaussian probability distributions are reflected.

Meanwhile, in setting the initial value by using Equation 1, the distribution modeling unit 130 may set an initial value by using a touch point within a preset range within a layout of a key instead of the entire touch point. Here, the touch point within the preset range in the layout of the key may refer to only a part of data selected in the order of distance from the center of the key among the touch points in the layout of the key. FIG. 1 is a diagram illustrating some data selected as an example.

After setting the initial value, the distribution modeling unit 130 estimates (learns) the parameters of the Gaussian mixture model using expectation maximization (EM). That is, in the first embodiment of the present invention, since there is no 'information about the touch point intended by the user' (which is missing), an expectation for the 'information about the touch point intended by the user' is obtained. The parameter is estimated (calculated) by repeating the process of calculating and maximizing the expected value. FIG. 4 illustrates a model of distribution for each key based on the estimated parameter.

Here, the expected value can be calculated using, for example, the Q function in Equation 2 below, where X means the entire touch point received and Y is the touch point intended by the user, which is missing information. Means information about.

In addition, estimating (calculating) the parameter through a process of maximizing the expected value may be expressed using, for example, Equation 3 below.

Meanwhile, since the expectation maximization process (EM) itself in Equations 2 and 3 is already known, a detailed description thereof will be omitted.

If the inverse covariance value is not estimated among the parameters, the distribution modeling unit 130 may estimate the inverse covariance value based on a pre-defined correction value. Epsilon

), Which forces the inverse covariance.

1 again, the key input interface device 100 according to the first embodiment of the present invention includes a setting unit 150. The setting unit 150 may set a recognition area recognized as an input for a key differently from the layout of the key, based on the distribution modeled by the distribution modeling unit 130.

To this end, the setting unit 150 may set a recognition area by selecting a parameter that maximizes likelihood for all keys based on the modeled distribution. That is, when selecting a parameter that maximizes the probability of each key, the boundary range for each key is calculated based on the point where the Gaussians for each key have the same height, and the boundary range sets the recognition area. 5 is a diagram exemplarily illustrating that a recognition area is set differently from the layout of a key as described above.

Therefore, according to an embodiment of the present invention, since the recognition area may be changed and set based on the distribution of the points touched by the user, the error rate may be reduced as much as possible.

In addition, when the input touch point is less than or equal to the preset number, the setting unit 150 may set the layout of the corresponding key as the recognition area or set the recognition area based on a previously modeled distribution. Therefore, even when there are keys with a small number of touch points which are the basis for learning the GMM using the EM, it is possible to set a recognition area for these keys.

In addition, in setting the recognition area, the setting unit 150 may be set only within a corresponding boundary value in consideration of whether a parameter is within a predetermined boundary. That is, for example, the setting unit 150 may allow the size of the recognition area or the distance from the center to exist within a predetermined boundary. By doing so, it is possible to prevent the recognition region from being mapped to a completely different key in learning GMM using EM.

In addition, the setting unit 150 may expand or reduce the recognition region of the key of the character predicted next based on the character string of the key input through the key input device. To this end, the key input interface device may further include a storage that stores a string of keys input through the key input device, and a predictor that predicts a character to be input next based on the stored string. The setting unit 150 may set a recognition region of a key corresponding to the character based on the predicted character.

Here, the prediction unit may predict the next character to be input by using a method such as N-gram, which is a well-known technique. When the N-gram counts N consecutive strings and then receives the N-1th character, This refers to a method of calculating the probability of a character to be input. For example, if the input string is 'abc', 'abc', or 'abd', the probability of entering 'c' when 'ab' is entered is twice that of 'd', and the N-gram itself is already Since it is a known technique, a detailed description thereof will be omitted.

In addition, the setting unit 150 adjusts an average value of the Gaussian parameters based on the probability predicted by the predictor to expand a recognition region of a key corresponding to the predicted character, and additionally recognizes a key around the key of the predicted character. Can be reduced.

As described above, according to the first exemplary embodiment of the present invention, the distribution of the point touched by the user is modeled as a probability model, and the error rate is reduced by adaptively changing the recognition region of the key based on the modeled distribution. You can.

FIG. 6 is a diagram illustrating a procedure of a method of interfacing a key input according to a first embodiment of the present invention. The method of interfacing the key input may be performed by the above-described key input interface device.

Referring to FIG. 6, in the method of interfacing a key input according to the first embodiment of the present invention, the method includes receiving a touch point (S110), modeling a distribution of a set of touch points (S130), and a recognition area. It includes the setting step (S150), but may further include other steps or may not include any one or more of these steps.

The receiving step S110 may be performed by the input unit 110 shown in FIG. 1, and receives a touch point touched by a user from a key input device. The touch point may be in the form of coordinates (x, y) on the touch screen, and as described above, the touch point may be input to the entire touch point from the key input device or may be input to some of the touch points.

The modeling step S130 may be performed by the distribution modeling unit 130 shown in FIG. 1, and the probability model defined in advance for each key includes a distribution of a set of touch points received in step S110. In this case, the distribution of the entire touch points, that is, the set of touch points, may be modeled.

The probabilistic model used for modeling in the step S130 of modeling is a parametric for estimating a parameter related to the probabilistic model, and models a distribution of a set of touch points without information about a key to be touched by a user. May be as described above. In addition, such a probability model may be, for example, a Gaussian mixture model (GMM), and the parameters of the Gaussian mixture model may be estimated (learned) using expectation maximization (EM). Same as one.

The modeling step (S130) may include setting an initial value. In this case, the initial value is set by using a touch point within a preset range within a layout of a key, not an entire touch point input. You can set the value.

The modeling step S130 may set an initial value and then estimate (learn) the parameters of the Gaussian mixture model using expectation maximization (EM). That is, in the first embodiment of the present invention, since there is no information about the touch point intended by the user (they are missing), an expectation for 'information about the touch point intended by the user' is calculated and such The process of maximizing the expected value is repeated, and the process of calculating the expected value may be expressed using Equation 2, wherein X denotes the entire touch point, and Y denotes the input point. As described above, the information about the touch point intended by the user, which is missing information, is meant.

In particular, in the step S130 of modeling according to the first embodiment of the present invention, when the inverse covariance value of the parameter is not estimated, the inverse covariance value may be estimated based on a predetermined correction value. In this case, the predefined correction value may be, for example, epsilon (_), thereby forcing the inverse covariance.

6 again, the setting of the recognition area (S150) may be performed by the setting unit 150 shown in FIG. 1, and the input for the key based on the distribution modeled in the modeling step (S130). A recognition area to be recognized is set, whereby the recognition area can be set (changed) differently from the layout of the key.

Therefore, since the recognition area can be changed and set based on the distribution of the points touched by the user, the error rate can be reduced as much as possible.

In addition, in the setting step (S150), when the input touch point is less than or equal to the preset number, the layout of the corresponding key may be set as the recognition area or the recognition area may be set based on a previously modeled distribution. Therefore, even when there are keys with a small number of touch points which are the basis for learning the GMM using the EM, it is possible to set a recognition area for these keys.

In addition, in the setting step (S150), in setting the recognition area, it may be set only within the corresponding boundary value in consideration of whether the parameter is within a predetermined boundary. That is, for example, the size or distance from the center of the recognition area may be present within a predetermined boundary, thereby preventing the recognition area from being mapped to a completely different key according to learning using EM.

In the setting step (S150), as described above, the key corresponding to the next character to be input is predicted based on the previously input character string, the recognition area corresponding to the predicted key is expanded, and the predicted key As described above, the recognition region of the surrounding key can be reduced.

Hereinafter, a key input interface device and method according to a second embodiment of the present invention will be described. However, since the second embodiment has a difference in that it further includes a labeling unit as compared with the first embodiment, the difference will be mainly described, and the same reference numerals and descriptions of the first embodiment will be used.

The key input interface device according to the second embodiment of the present invention further includes a labeling unit in the key input interface device 100 shown in FIG. 1, and the key input interface device further including the labeling unit will be described in detail below. Let's look at it.

First, since the operations and effects of the input unit 110 and the distribution modeling unit 130 are the same as those of the first embodiment, detailed description thereof will be omitted.

The labeling unit may select a hit based on a distribution modeled by the distribution modeling unit 130 when the likelihood calculated by the distribution modeling unit 130 with respect to the input touch point is equal to or greater than a preset threshold. In addition, the labeling unit may select a typo from the touch point using a predefined typo detection algorithm. For example, a known technique such as backtracking of a backspace may be applied to the typo detection algorithm.

The setting unit 150 sets a recognition area recognized as an input to a key based on the labeling result. At this time, the recognition area may be set using a deep neural network. .

According to the key input interface device according to the second embodiment of the present invention having the above-described configuration, labeling may be performed based on whether an input touch point is above a certain reliability level and is a typo. This allows more precise modeling of the distribution of touch points.

Hereinafter, a key input interface device and method according to a third embodiment of the present invention will be described. However, since the third embodiment has a difference in that it further includes a situation determination unit in comparison with the first embodiment and the second embodiment, the difference will be described mainly, and the same parts will be described in the first and second embodiments. Description and reference numerals are used.

The key input interface device according to the third embodiment of the present invention further includes a situation determination unit in the key input interface device 100 illustrated in FIGS. 1 and 2, and further includes a situation determination unit below. Let's take a closer look at.

The situation determination unit determines the situation of the user of the key input device. To this end, the situation determination unit may include, for example, a gyro sensor, an acceleration sensor, an illumination sensor, a gps sensor, a temperature or humidity sensor, an image recognition sensor, or a voice sensor, and the like based on information collected from the sensors. You can judge the situation. For example, the situation determination unit may determine that the user is lying down by using the gyro sensor when it is inclined at a predetermined angle or more, and using the temperature sensor, whether the temperature where the user uses the key input device is cold or It may determine whether it is hot or the like, and may further determine whether the user's posture, whether the user manipulates the key input device with both hands, and the like. In this case, the situation determination unit may determine a situation using a decision tree with respect to information collected from a sensor, but is not limited thereto.

According to the key input interface device according to the third embodiment of the present invention having the above-described configuration, the input unit 110 may classify the user's situation of the input touch point according to the situation determined by the situation determination unit. The modeling unit 130 models the distribution differently according to the situation determined by the situation determination unit. In addition, the setting unit 150 sets the recognition area differently according to the situation.

Therefore, according to the third exemplary embodiment of the present invention, the recognition area most suitable for the user's situation can be provided, thereby reducing the error rate.

In addition, the method according to an embodiment of the present invention may be implemented in a recording medium having a computer program recorded thereon.

Combinations of each block of the block diagrams and respective steps of the flowcharts attached to the present invention may be performed by computer program instructions. These computer program instructions may be mounted on a processor of a general purpose computer, special purpose computer, or other programmable data processing equipment such that instructions executed through the processor of the computer or other programmable data processing equipment may not be included in each block or flowchart of the block diagram. It will create means for performing the functions described in each step. These computer program instructions may be stored in a computer usable or computer readable memory that can be directed to a computer or other programmable data processing equipment to implement functionality in a particular manner, and thus the computer usable or computer readable memory. It is also possible for the instructions stored in to produce an article of manufacture containing instruction means for performing the functions described in each block or flowchart of each step of the block diagram. Computer program instructions may also be mounted on a computer or other programmable data processing equipment, such that a series of operating steps may be performed on the computer or other programmable data processing equipment to create a computer-implemented process to create a computer or other programmable data. Instructions that perform processing equipment may also provide steps for performing the functions described in each block of the block diagram and in each step of the flowchart.

In addition, each block or step may represent a portion of a module, segment or code that includes one or more executable instructions for executing a specified logical function (s). It should also be noted that in some alternative embodiments, the functions noted in the blocks or steps may occur out of order. For example, the two blocks or steps shown in succession may in fact be executed substantially concurrently or the blocks or steps may sometimes be performed in the reverse order, depending on the functionality involved.

The above description is merely illustrative of the technical idea of the present invention, and those skilled in the art to which the present invention pertains may make various modifications and changes without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention but to describe the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas falling within the scope of the present invention should be construed as being included in the scope of the present invention.

Claims

An input unit configured to receive touch points for a plurality of keys included in the key input device;

A distribution modeling unit for modeling a distribution of the set of touch points with respect to each of the keys by using a predefined probability model; And

And a setting unit configured to set a recognition area recognized as an input to the key, based on the modeled distribution.

Key in interface device.
The method of claim 1,

The probability model is

Characterized in that the key intended by the touch point is unsupervised learning on information about which key among the plurality of keys.

Key in interface device.
The method of claim 1,

The probability model is

Characterized in that it is a parametric for estimating a parameter associated with the probability model

Key in interface device.
The method of claim 3, wherein

The distribution modeling unit,

Setting an initial value of the parameter based on a touch point within a preset range within the layout of the key

Key in interface device.
The method of claim 3, wherein

The distribution modeling unit,

If an inverse covariance value is not estimated among the parameters, the inverse covariance value is estimated based on a predetermined correction value.

Key in interface device.
The method of claim 1,

The probability model is a Gaussian mixture model (GMM),

The distribution modeling unit,

Estimate the parameters of the Gaussian mixture model using expectation maximization (EM)

Key in interface device.
The method of claim 1,

The setting unit,

The recognition region is set based on a first layout corresponding to the first key or the recognition is performed based on a first distribution previously modeled for the first key having the touch points input from the plurality of keys equal to or less than a preset number. To set the area

Key in interface device.
The method of claim 1,

The key input interface device,

A storage unit which stores a character string of a key input through the key input device; And

Further comprising a prediction unit for predicting the next character to be input based on the stored string,

The setting unit,

Setting a recognition region of a key corresponding to the character based on the predicted character

Key in interface device.
The method of claim 8,

The setting unit,

The recognition region of the key corresponding to the predicted character is expanded, and the recognition region of the key around the key corresponding to the predicted character is reduced.

Key in interface device.
An input unit configured to receive touch points for a plurality of keys included in the key input device;

A distribution modeling unit modeling a distribution of the set of touch points for each of the keys by using a predefined probability model;

A labeling unit for classifying and labeling hits and typos among the input touch points based on the modeled distribution; And

And a setting unit configured to set a recognition area recognized as an input to the key, based on the labeling result.

Key in interface device.
The method of claim 10,

The labeling unit,

If the likelihood of the input touch point is greater than or equal to a preset threshold, screening is performed with a fixed stroke.

Key in interface device.
The method of claim 10,

The labeling unit,

Selecting a typo from the input touch point based on a predefined typo detection algorithm

Key in interface device.
The method of claim 10,

The setting unit,

Based on the labeling result, the recognition region is set using a deep neural network.

Key in interface device.
The method of claim 1,

The key input interface device further includes a situation determination unit that determines a situation of a user of the key input device.

The input unit classifies the input touch point according to the situation,

The distribution modeling unit models the distribution differently according to the situation,

The setting unit sets the recognition area differently according to the situation.

Key in interface device.
The method of claim 14,

The above situation,

At least one of a movement of the user, a posture of the user, whether the user operates the key input device with both hands, or a temperature at which the key input device is operated.

Key in interface device.
A method of interfacing a key input of a key input device performed by a key input interface device,

Receiving a touch point for a plurality of keys included in the key input device;

Modeling a distribution of the set of touch points for each key using a predefined probability model; And

Setting a recognition region recognized as an input to the key, based on the modeled distribution;

Keystroke interface method.
The method of claim 16,

The probability model is a Gaussian mixture model (GMM),

The modeling step,

Estimate the parameters of the Gaussian mixture model using expectation maximization (EM)

Keystroke interface method.
The method of claim 16,

The key input interface method further includes determining a situation of a user of the key input device.

In the receiving step, the received touch point is classified according to the situation.

The modeling may include differently modeling the distribution according to the situation.

The setting may include setting the recognition area differently according to the situation.

Keystroke interface method.