Electronic device for identifying abnormal touch and control method thereof

Abstract

Disclosed is an electronic device. The electronic device comprises: a memory for storing a first artificial intelligence model for outputting touch information on the basis of touch data input during the current cycle and a second artificial intelligence model for outputting touch information on the basis of the touch data input during the current cycle and previous state information; a touch panel; a sensor; and one or more processors, wherein the one or more processors can input touch data obtained through the touch panel to the first artificial intelligence model and the second artificial intelligence model, and if the touch information output from the first artificial intelligence model and the second artificial intelligence model are different, perform an operation corresponding to the touch information output from the first artificial intelligence model.

Description

Electronic device for identifying abnormal touch and method for controlling the same The present invention relates to an electronic device and a control method thereof, and more particularly, to an electronic device capable of identifying an abnormal touch and a control method thereof. Recently, electronic devices that receive user input via touch panels and perform corresponding actions are becoming more widespread. However, unintended touch inputs on the touch panel have led to problems such as unintentional calls to other people or the launch of applications. Conventional technology has used various sensors, such as proximity sensors and light sensors, to identify whether touch input via a touch panel is normal. For example, if a sensor detects that an electronic device is consistently placed close to an external object or in a dark location, the electronic device is assumed to be in a pocket, and the touch input to the touch panel is considered an abnormal touch input from inside the pocket. However, relying solely on the sensor's sensing values to determine whether a touch is normal has low accuracy. Furthermore, the physical space occupied by various sensors increases, leaving less room for memory and processors. Furthermore, the high cost of sensors increases the manufacturing cost of electronic devices. Therefore, to reduce the physical space occupied by sensors and manufacturing costs, there has been a growing need for technology that can accurately determine whether a touch is normal using only a small number of low-cost sensors. An electronic device according to at least one embodiment of the present disclosure includes a memory storing a first artificial intelligence model for outputting touch information based on touch data input in a current period and a second artificial intelligence model for outputting touch information based on the touch data input in the current period and previous state information, a touch panel, a sensor, and one or more processors, wherein the one or more processors input touch data acquired through the touch panel into the first artificial intelligence model and the second artificial intelligence model, and when the touch information output from the first artificial intelligence model and the second artificial intelligence model is different, perform an operation corresponding to the touch information output from the first artificial intelligence model. A method for controlling an electronic device according to at least one embodiment of the present disclosure may include a step of inputting acquired touch data into a first artificial intelligence model and a second artificial intelligence model, and a step of performing an operation corresponding to the touch information output from the first artificial intelligence model if the touch information output from the first artificial intelligence model and the second artificial intelligence model are different. A computer-readable recording medium including a program for executing a control method of an electronic device according to at least one embodiment of the present disclosure, wherein the control method of the electronic device may include a step of inputting acquired touch data into a first artificial intelligence model and a second artificial intelligence model, and a step of performing an operation corresponding to the touch information output from the first artificial intelligence model if the touch information output from the first artificial intelligence model and the second artificial intelligence model are different. FIG. 1 is a diagram illustrating the operation of an electronic device according to one or more embodiments of the present disclosure. FIG. 2 is a block diagram illustrating a configuration of an electronic device according to one or more embodiments of the present disclosure. FIG. 3 is a diagram illustrating a method for identifying whether a touch is normal by a first artificial intelligence model according to one or more embodiments of the present disclosure. FIG. 4 is a diagram for explaining a method for identifying whether a touch is normal by a second artificial intelligence model according to one or more embodiments of the present disclosure. FIG. 5 is a diagram illustrating a method for determining whether to initialize the state of a second artificial intelligence model of an electronic device according to one or more embodiments of the present disclosure. FIG. 6 is a diagram illustrating a method for calculating an initialization score for initializing the state of a second artificial intelligence model of an electronic device according to one or more embodiments of the present disclosure. FIG. 7 is a diagram illustrating a method for initializing a state of a second artificial intelligence model according to one or more embodiments of the present disclosure. FIG. 8 is a diagram illustrating a data input method for a first artificial intelligence model and a second artificial intelligence model according to one or more embodiments of the present disclosure. FIG. 9 is a flowchart illustrating a method for controlling an electronic device according to one or more embodiments of the present disclosure. FIG. 10 is a flowchart illustrating a method for controlling an electronic device according to one or more embodiments of the present disclosure. The present embodiments may be modified and have various embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the scope to specific embodiments, but should be understood to encompass various modifications, equivalents, and / or alternatives of the embodiments of the present disclosure. In connection with the description of the drawings, similar reference numerals may be used for similar components. In describing the present disclosure, if it is determined that a specific description of a related known function or configuration may unnecessarily obscure the gist of the present disclosure, a detailed description thereof will be omitted. Additionally, the following embodiments may be modified in various other forms, and the scope of the technical concepts of the present disclosure is not limited to the following embodiments. Rather, these embodiments are provided to further faithfully and completely convey the technical concepts of the present disclosure to those skilled in the art. The terminology used in this disclosure is for the purpose of describing specific embodiments only and is not intended to limit the scope of the rights. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this disclosure, expressions such as “has,” “can have,” “includes,” or “may include” indicate the presence of a corresponding feature (e.g., a component such as a number, function, operation, or part), and do not exclude the presence of additional features. In this disclosure, expressions such as “A or B,” “at least one of A and / or B,” or “one or more of A or / and B” can include all possible combinations of the listed items. For example, “A or B,” “at least one of A and B,” or “at least one of A or B” can all refer to (1) including at least one A, (2) including at least one B, or (3) including both at least one A and at least one B. The expressions “first,” “second,” “first,” or “second,” etc., used in this disclosure can describe various components, regardless of order and / or importance, and are only used to distinguish one component from another, but do not limit the components. When it is said that a component (e.g., a first component) is “(operatively or communicatively) coupled with / to” or “connected to” another component (e.g., a second component), it should be understood that said component may be directly coupled to said other component, or may be coupled via another component (e.g., a third component). On the other hand, when it is said that a component (e.g., a first component) is "directly connected" or "directly connected" to another component (e.g., a second component), it can be understood that no other component (e.g., a third component) exists between said component and said other component. The expression "configured to" as used in the present disclosure may be used interchangeably with, for example, "suitable for," "having the capacity to," "designed to," "adapted to," "made to," or "capable of." The term "configured to" may not necessarily mean only "specifically designed to" in terms of hardware. Instead, in some contexts, the phrase "a device configured to" may mean that the device, in conjunction with other devices or components, is "capable of" performing A, B, and C. For example, the phrase "a processor configured (or set) to perform A, B, and C" may refer to a dedicated processor (e.g., an embedded processor) for performing those operations, or a general-purpose processor (e.g., a CPU or application processor) that can perform those operations by executing one or more software programs stored in a memory device. In the embodiments, a 'module' or 'part' performs at least one function or operation, and may be implemented as hardware or software, or as a combination of hardware and software. Furthermore, a plurality of 'modules' or 'parts' may be integrated into at least one module and implemented as at least one processor, except for a 'module' or 'part' that needs to be implemented as a specific hardware. Meanwhile, the various elements and areas in the drawings are schematically drawn. Therefore, the technical concept of the present invention is not limited by the relative sizes or spacing depicted in the attached drawings. Hereinafter, with reference to the attached drawings, embodiments according to the present disclosure will be described in detail so that a person having ordinary knowledge in the technical field to which the present disclosure pertains can easily implement the present disclosure. FIG. 1 is a diagram illustrating the operation of an electronic device according to one or more embodiments of the present disclosure. According to FIG. 1, if a touch input through a touch panel is an abnormal touch, the electronic device (100) can prevent an operation of the electronic device (100) based on an abnormal touch by not performing any action or by receiving a user's confirmation. The electronic device (100) may be a device that includes a touch panel capable of receiving a user's touch input, and performs an operation corresponding to a touch input through the touch panel. In FIG. 1, the electronic device (100) is illustrated as a mobile phone, but this is only one example, and may be implemented as various types of electronic devices capable of touch input, such as an electronic picture frame, a DID (Digital Information Display), a kiosk, a PMP (Portable Media Layer), an MP3 player, a game console, a LFD (Large Format Display), a laptop, a TV, a tablet PC, a monitor, a projector system, etc. However, the present invention is not limited thereto, and other home appliances, medical devices, etc. may also be included in the electronic device (100). A touch panel is a component capable of detecting a touch, and may be implemented as a touch screen capable of displaying, or may be implemented in various forms, such as a touch pad or a touch button. According to one embodiment of the present disclosure, the electronic device (100) can identify whether a touch input through the touch panel corresponds to an abnormal touch or a normal touch based on an output value of an artificial intelligence model. Unlike existing technologies that identify whether a touch is normal through various types of sensors, the electronic device (100) of the present disclosure can identify whether a touch is normal based on an output value of an artificial intelligence model, thereby securing the physical space occupied by various types of sensors within the electronic device (100) and reducing manufacturing costs by eliminating the need for expensive sensors. In addition, the electronic device (100) can more accurately identify whether a touch is normal by using a plurality of artificial intelligence models. A specific method for identifying whether a touch is normal will be described in detail in the description of the drawings to be described later. Here, a “normal touch” may mean that a user inputs a touch on the touch panel using a touch object such as a finger, a touch pen, a stylus pen, a pointer, etc. For example, a “normal touch” may include a user contacting a fingerprint of a finger on the touch panel to unlock the electronic device (100). “Normal touch” may be replaced with other terms such as “finger touch, pen touch, pointer touch, intentional touch, non-accidental touch, etc.”, but in this specification, the term “normal touch” is used interchangeably. Here, an "abnormal touch" may be a touch that is not intended by the user. For example, as illustrated in FIG. 1, an "abnormal touch" may include a touch input to the touch panel of the electronic device (100) by an object inside the pocket, etc., when the user puts the electronic device (100) in the pocket. Alternatively, when a user carries an electronic device in a bag or carries it in his or her hand, a finger, a touch pen, or other external object may touch the touch panel even when the user does not intend to operate the electronic device. Alternatively, in the case of an electronic device with a large screen, such as a TV, animals such as dogs, cats, or birds may touch the TV screen, or an object thrown by the user may touch the screen. Most of these touches occur unintentionally. In this disclosure, these situations are referred to as abnormal touches. “Abnormal touch” may be replaced with other terms such as “in-pocket touch,” “object touch,” “non-intentional touch,” and “accidental touch,” but in this specification, the term “abnormal touch” is used interchangeably. In FIG. 1, it is illustrated that the electronic device (100) executes a malfunction prevention filter based on the identification of an abnormal touch to control the electronic device (100) not to operate due to an abnormal touch. According to FIG. 1, a user can recognize that a touch has been made by looking at the screen of the electronic device (100). If the touch is intended by the user, i.e., a normal touch, the user can touch an object displayed on the screen and drag it in the upward arrow direction. In this case, the electronic device (100) can unlock and display a standby screen. On the other hand, if the touch is not intended by the user, the user can do nothing or press a button to switch back to a sleep state. If there is no operation for a certain period of time, the electronic device (100) can switch to a sleep state and deactivate the screen. Although FIG. 1 illustrates a case where a malfunction prevention filter is provided, the electronic device (100) is not limited thereto, and may display a UI screen indicating that an abnormal touch has been input. Alternatively, the electronic device (100) may generate vibrations to make the user recognize that a touch has been made, or may ignore the touch and not turn on the power or perform any other action. Meanwhile, if the electronic device (100) determines that a touch is normal, it can perform an action corresponding to the touch information. For example, if a plurality of application icons are displayed on a touch screen and a point on the touch screen is normally touched, the electronic device (100) can identify the touch coordinates and execute the application corresponding to the icon displayed at the touch coordinates. The electronic device (100) can sense the situation before and after a touch and determine whether the touch is normal or abnormal based on that. The specific determination method will be described in detail later. Below, the operation of an electronic device (100) according to various embodiments of the present disclosure will be described. FIG. 2 is a block diagram illustrating a configuration of an electronic device according to one or more embodiments of the present disclosure. According to FIG. 2, the electronic device (100) may include a memory (110), a touch panel (120), a sensor (130), and a processor (140). As described above, the electronic device (100) may be implemented in various forms of devices, but since abnormal touch is more likely to occur in a small electronic device that a user can carry around in a bag or pocket, the following description will be based on a case implemented in the form of a mobile phone as shown in FIG. 1. The memory (110) can store various programs, data, commands, etc. used in the electronic device (100). Furthermore, the memory (110) can store a first artificial intelligence model and a second artificial intelligence model. Furthermore, the memory (110) can store various pieces of information according to various embodiments of the present disclosure. Here, the "first artificial intelligence model" may be an artificial intelligence model trained to receive touch data acquired through a touch panel and output information on whether the touch data corresponds to a normal touch. The first artificial intelligence model may receive touch data at preset intervals, generate current status information based on the touch data input in the current cycle, and output information on whether the touch data input in the current cycle corresponds to a normal touch based on the current status information. Here, unlike the first artificial intelligence model, the "second artificial intelligence model" can output information on whether the touch data corresponds to a normal touch based on previous state information and the touch data input in the current cycle. Therefore, the second artificial intelligence model can output information based on the previous state information, and thus can output information by reflecting the characteristics of not only the touch data input in the current cycle but also the touch data input in the previous cycle. Accordingly, the second artificial intelligence model may be an artificial intelligence model suitable for outputting information on continuous touch data. The description of the first artificial intelligence model and the second artificial intelligence model will be described in detail in the description of FIGS. 3 to 8 described below. A memory (110) according to an example of the present disclosure may be implemented as an internal memory such as a ROM (e.g., an electrically erasable programmable read-only memory (EEPROM)) or a RAM included in one or more processors (140), or may be implemented as a separate memory from one or more processors (140). In this case, the memory (110) may be implemented as a memory embedded in the electronic device (100) or as a memory detachable from the electronic device (100) depending on the purpose of data storage. For example, data for driving the electronic device (100) may be stored in a memory embedded in the electronic device (100), and data for an extended function of the electronic device (100) may be stored in a memory detachable from the electronic device (100). Meanwhile, in the case of memory embedded in the electronic device (100), it may be implemented as at least one of volatile memory (e.g., dynamic RAM (DRAM), static RAM (SRAM), or synchronous dynamic RAM (SDRAM)), non-volatile memory (e.g., one time programmable ROM (OTPROM), programmable ROM (PROM), erasable and programmable ROM (EPROM), electrically erasable and programmable ROM (EEPROM), mask ROM, flash ROM, flash memory (e.g., NAND flash or NOR flash), hard drive, or solid state drive (SSD)), and in the case of memory that can be detachably attached to the electronic device (100), it may be implemented as a memory card (e.g., compact flash (CF), secure digital (SD), micro secure digital (Micro-SD), mini secure digital (Mini-SD), extreme digital (xD), multi-media card (MMC), etc.), external memory that can be connected to a USB port (e.g., USB memory), etc. there is. The touch panel (120) is a component for sensing touch and outputting touch data. As described above, the touch panel (120) may be implemented in various forms, such as a touch screen or a touch button. In the present disclosure, the term "touch" may include not only direct contact of a conductive object on the touch panel (120) of the electronic device (100) but also proximity of a conductive object to the screen of the electronic device (100). For example, the conductive object may include a user's body (e.g., a finger, a palm, a face, an ear, a thigh, a buttocks, etc.), a touch pen, a stylus pen, etc. Alternatively, the touch panel may be a glove with a conductive pad attached that connects the user's hand to the outside world. The touch panel (120) may be embedded in the display panel of the electronic device (100) or may be provided separately from the display panel. For example, when implemented as an electronic device (100) having a separate touch panel (120) provided around a display panel, a user may perform an operation such as moving a cursor displayed on the display panel by touching and dragging a point on the touch panel (120). The touch panel (120) may include a plurality of sensing electrodes arranged in rows and columns. Touch data may be output through the plurality of sensing electrodes. For example, the plurality of sensing electrodes may output touch data according to a capacitance sensing method. According to one embodiment, the touch panel (120) may be implemented with a structure in which capacitance is formed between each of the plurality of sensing electrodes and the ground, and such a structure may be referred to as a self-capacitance structure. In this case, the touch panel (120) may detect the amount of capacitance change of each of the plurality of sensing electrodes and output touch data in a self-capacitance manner. According to another embodiment, the touch panel (120) may be implemented with a structure in which capacitance is formed between a plurality of sensing electrodes, and such a structure may be referred to as a mutual capacitance structure. The touch panel (120) implemented with a mutual capacitance structure may include a sensing electrode layer arranged in a horizontal axis and a sensing electrode layer arranged in a vertical axis, and a capacitance may be formed between the electrode layers at the intersection of the horizontal axis and the vertical axis. In this case, the touch panel (120) may detect a change in the capacitance formed at the intersection and output touch data in a mutual capacitance manner. In the above description, it was explained that the touch panel (120) is implemented with a self-capacitance structure and a mutual capacitance structure and can output touch data of the self-capacitance method and touch data of the mutual capacitance method, respectively. However, it is of course possible for the touch panel (120) to simultaneously output touch data of the self-capacitance method and touch data of the mutual capacitance method. In addition, although the above description only explains that the touch panel (120) can output touch data according to the electrostatic capacitance sensing method, this is only one example, and it is of course possible to sense touch and output touch data according to various sensing methods such as resistive sensing, optical sensing, and pressure sensing. However, in this description, for the convenience of explanation, only the touch panel (120) using the electrostatic capacitance sensing method will be described. The sensor (130) is configured to detect a change in the status of the electronic device (100). According to one embodiment, the sensor (130) may include an illuminance sensor. The illuminance sensor may convert light incident on the electronic device (100) into an electrical signal and output an electrical signal representing the density of the incident light. The processor (140) may identify whether the electronic device (100) is located in a bright or dark place based on the electrical signal output by the illuminance sensor. According to another embodiment, the sensor (130) may include a motion detect sensor such as a gyro sensor, an acceleration sensor, a geomagnetic sensor, etc. Here, the motion detect sensor may include various sensors that detect movement, rotation, inclination, acceleration, etc. of the electronic device (100). The motion detect sensor may detect motion such as movement or rotation of the electronic device (100) and output an electrical signal corresponding to the motion. The processor (140) may identify how much the electronic device (100) has moved, in which direction and how much it has rotated, etc., based on the electrical signal output by the motion detecting sensor. For example, the processor (140) may identify changes in pitch angle, roll angle, yaw angle, etc. In the above description, it is only exemplified that the sensor (130) may include a light sensor or a motion detection sensor, but this is only one example, and it is of course possible to include various sensors such as a proximity sensor, an ultrasonic sensor, an infrared sensor, etc., through which a change in the state of the electronic device (100) can be detected. For example, when the sensor (130) includes a proximity sensor, the processor (140) can identify that the electronic device (100) is continuously in proximity to an external object through a sensing value of the proximity sensor. One or more processors (140) control the overall operation of the electronic device (100). Specifically, one or more processors (140) may be connected to each component of the electronic device (100) to control the overall operation of the electronic device (100). For example, one or more processors (140) may be operatively connected to a memory (110), a touch panel (120), and a sensor (130). The processor (140) may be composed of one or more processors. One or more processors (140) may perform operations of the electronic device (100) according to various embodiments by executing at least one instruction stored in the memory (110). The one or more processors (140) may include one or more of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), an Accelerated Processing Unit (APU), a Many Integrated Core (MIC), a Digital Signal Processor (DSP), a Neural Processing Unit (NPU), a hardware accelerator, or a machine learning accelerator. The one or more processors (140) may control one or any combination of other components of the electronic device, and may perform operations related to communication or data processing. The one or more processors (140) may execute one or more programs or instructions stored in a memory. For example, the one or more processors may perform a method according to one or more embodiments of the present disclosure by executing one or more instructions stored in a memory. One or more processors (140) may be implemented as dedicated AI processors. For example, one or more processors (140) may be designed as hardware chips, such as ASICs or FPGAs, specialized for processing specific AI models. As described above, when at least one AI model is stored in the memory (110), the processor (140) may execute the AI model to perform various operations. Specifically, the processor (140) can input touch data acquired through the touch panel (120) into the first artificial intelligence model and the second artificial intelligence model. The first artificial intelligence model and the second artificial intelligence model can output touch information according to the execution of the processor (140). Here, the touch data acquired through the touch panel (120) may be in the form of a two-dimensional array. Specifically, the touch panel (120) may be divided into a plurality of cells in the horizontal and vertical directions, and the touch data may be data in the form of a two-dimensional array that records the amount of change in electrostatic capacity that occurs in each of the plurality of cells due to a touch input. For example, a change in electrostatic capacity of 100 pF may be detected in the cell located 10th from the right and 5th from the bottom by a touch input to the touch panel (120), and in this case, the array

[0010] [5] = 100 (pF) can be stored as a two-dimensional array of touch data. At this time, the greater the intensity of the touch input to the touch panel (120), the greater the change in electrostatic capacity. Accordingly, the two-dimensional array of touch data can include information on the coordinates where the touch was input and information on the intensity of the touch. Here, the touch information may include information indicating whether the touch input through the touch panel (120) is a normal touch or an abnormal touch. For example, the first artificial intelligence model and the second artificial intelligence model may output a probability that the touch data input to the model corresponds to a normal touch and a probability that the touch data corresponds to an abnormal touch. If the probability of a normal touch is greater, the processor (140) may identify the input touch data as a normal touch. On the other hand, if the probability of an abnormal touch is greater, the processor (140) may identify the input touch data as an abnormal touch. In the above description, it is only exemplified that the touch data acquired through the touch panel (120) is in the form of a two-dimensional array, but this is only one example, and it is obvious that the touch data can be implemented in the form of a three-dimensional array or a multi-dimensional array. In the above description, it is only exemplified that touch information can mean information about whether a touch is normal or abnormal, but this is only one example, and touch information can include information about a specific body part that generated a touch input, such as whether it is a touch by the face or a touch by the thigh, and of course, it can include various specific information depending on the learning method of the artificial intelligence model. The touch information output by the first and second AI models can be identical in most cases. However, if a sudden change in circumstances occurs, such as when the user puts the electronic device (100) in and out of their pocket, or vice versa, suddenly puts it in, there is a possibility that one of the two models will output incorrect touch information. According to one embodiment, if the touch information output from the first artificial intelligence model and the second artificial intelligence model are different, the processor (140) may perform an operation corresponding to the touch information output from the first artificial intelligence model. According to another embodiment, if the touch information output from the first artificial intelligence model and the second artificial intelligence model are different, the processor (140) can identify the amount of change in the sensing value of the sensor (130) for a preset unit time prior to a touch on the touch panel. Since the first artificial intelligence model outputs touch information based on the touch data input in the current cycle, while the second artificial intelligence model outputs touch information based on the touch data input in the current cycle and previous state information, the touch information output by the two artificial intelligence models may be different. Here, the term "cycle" may refer to a cycle in which the first and second artificial intelligence models output touch information or a cycle in which touch data is input to the first and second artificial intelligence models. For example, if the processor (140) acquires touch data corresponding to the user's touch input to the touch panel (120) at a frequency of 120 Hz, the touch data may be input to the first and second artificial intelligence models. can be input in cycles. In this case, the first and second artificial intelligence models Touch information can be output in cycles. “Cycle” can be replaced with terms such as stage, time step, and state, but in this description, the term “cycle” is used interchangeably. In addition, "state information" may mean information generated by the first and second artificial intelligence models to output touch information, and the "state information" may include features of touch data input in the current cycle (e.g., area where touch is input, strength of touch) and features of touch data input in the previous cycle. In this case, the first artificial intelligence model may generate current state information based on features extracted from touch data input in the current cycle, and the second artificial intelligence model may generate current state information based on features extracted from touch data input in the current cycle and previous state information. According to one embodiment, If the cycle is set to , the previous state information is used by the second artificial intelligence model in the current cycle. It may mean state information generated based on features extracted from previously acquired touch data. In this case, the previous state information It may include features extracted from previously acquired touch data. Since the second artificial intelligence model generates current state information based on previous state information and touch data input in the current cycle, and outputs touch information based on the current state information, it can output touch information that reflects the features of touch data input in the previous cycle, and thus may be suitable for outputting touch information for continuously input touch data. According to one embodiment, as illustrated in FIG. 1, when the electronic device (100) is positioned inside a pocket, a continuous abnormal touch may be input to the touch panel (120). In this state, when a user wishes to use the electronic device (100), the user may take the electronic device (100) out of the pocket and touch the touch panel (120) with a finger. In this way, even if a normal touch by the user's finger is input to the touch panel (120) after suddenly moving from inside to outside the pocket, there is a possibility that the second artificial intelligence model will identify it as an abnormal touch. This is because the second artificial intelligence model outputs touch information based on not only the touch data input in the current cycle but also previous state information, and the previous state information includes the characteristics of the abnormal touches input to the touch panel (120) when the electronic device (100) is positioned inside the pocket. As in the situation described above, if an abnormal touch is continuously input to the touch panel (120) in the previous cycle and a normal touch is input to the touch panel (120) in the current cycle, the first artificial intelligence model may output touch information corresponding to the normal touch, but the second artificial intelligence model may output different touch information corresponding to the abnormal touch. In this case, the processor (140) can identify whether the amount of change in the sensing value of the sensor (130) during the preset unit time prior to the touch on the touch panel (120) in the current cycle exceeds the threshold value. According to one embodiment, when the sensor (130) includes a light sensor, it is possible to identify whether the sensing value of the light sensor has increased by an amount exceeding a threshold value during a preset unit time prior to a touch on the touch panel (120) in the current cycle. Specifically, when the sensing value of the light sensor changes from indicating 0.01 lux to indicating 10 lux within 3 seconds, the processor (140) can identify that the sensing value has changed by an amount exceeding the threshold value, and through this, it can also identify that the electronic device (100) has been moved from being inside a pocket to outside. According to another embodiment, the sensor (130) may include a motion detection sensor, and the processor (140) may identify whether the amount of change in the sensing value of the motion detection sensor exceeds a threshold value. Specifically, if the amount of change in rotational momentum and the amount of change in linear momentum detected by the motion detection sensor exceed the threshold value, the processor (140) may identify that the electronic device (100) has moved from inside the pocket to outside the pocket, or has moved from outside the pocket to inside the pocket. In the above description, the “preset unit time” is exemplified as 3 seconds, but this is only an example, and the “preset unit time” can be set to any time, such as 1 second, 2 seconds, etc., and can also be set by various criteria, such as the average time required to move the electronic device (100) from inside the pocket to outside, or the average time required to move the electronic device (100) from inside the bag to outside. The processor (140) can perform an operation corresponding to touch information output from the first artificial intelligence model when the amount of change in the sensing value exceeds the threshold value, and can perform an operation corresponding to touch information output from the second artificial intelligence model when the amount of change in the sensing value is less than or equal to the threshold value. According to one embodiment, if the touch information output from the first artificial intelligence model or the second artificial intelligence model corresponds to a normal touch, the processor (140) may perform an operation corresponding to the normal touch. For example, the processor (140) may perform various operations based on the coordinates where the normal touch is detected on the touch panel (120), such as turning on the power of the electronic device (100), unlocking it, or executing / terminating an application. According to another embodiment, if the touch information output from the first artificial intelligence model or the second artificial intelligence model corresponds to an abnormal touch, the processor (140) may perform an operation corresponding to the abnormal touch. For example, the processor (140) may perform various operations, such as not turning on the power of the electronic device (100), displaying a malfunction prevention filter, or displaying a UI screen indicating that an abnormal touch is being input. In addition, the processor (140) may initialize the state of the second artificial intelligence model when the amount of change in the sensing value exceeds the threshold value, which will be described in detail in the description of the drawings to be described later. As described above, normal touch includes various types of touches input to the touch panel (120) using a touch object such as a finger, a touch pen, a stylus pen, a pointer, etc., and abnormal touch may include various types of touches input to the touch panel (120) without using a touch object. However, in the description of the drawings described below, for the convenience of explanation, only finger touch and pocket touch will be described as examples of normal touch and abnormal touch. FIG. 3 is a drawing for explaining a method for identifying whether a touch is normal by a first artificial intelligence model according to one or more embodiments of the present disclosure, and FIG. 4 is a drawing for explaining a method for identifying whether a touch is normal by a second artificial intelligence model according to one or more embodiments of the present disclosure. As illustrated in FIG. 3, the processor (140) can input touch data (301, 302, 303, 304, 305, 306) to the first artificial intelligence model (210) and output touch information (307, 308, 309, 310, 311, 312). For example, the processor (140) may input touch data (301) corresponding to a pocket touch into the first artificial intelligence model (210) to output touch information (307) indicating an abnormal touch. At this time, the first artificial intelligence model (210) may not be suitable for outputting touch information for touch data that is continuously input. For example, let's assume that a pocket touch is input three times in succession. The processor (140) may input touch data (303) corresponding to the third input pocket touch into the first artificial intelligence model (210) to obtain incorrect touch information (309) as a normal touch. The reason why incorrect touch information is output from the first artificial intelligence model (210) may be that the parameters of the first artificial intelligence model (210) are not properly adjusted due to insufficient learning data of the first artificial intelligence model (210), or that the pocket touch data (303) contains some features similar to finger touch data. However, the first artificial intelligence model (210) can output accurate touch information even when the touch data continuously acquired through the touch panel (120) is converted from pocket touch data (301, 302, 303) to finger touch data (304). For example, even if pocket touch data is inputted three times consecutively to the first artificial intelligence model (210) and then finger touch data (304) is inputted, the first artificial intelligence model (210) can output touch information (310) indicating a normal touch. Since the first artificial intelligence model (210) is not affected by previous state information and outputs touch information based only on the currently input touch data, it can output more accurate touch information than the second artificial intelligence model (220) even when the type of continuously input touch data changes. Meanwhile, as illustrated in FIG. 4, the processor (140) can input touch data (301, 302, 303, 304, 305, 306) to the second artificial intelligence model (220) to obtain touch information (407, 408, 409, 410, 411, 412). Since the processor (140) controls the state information generated by the second artificial intelligence model (220) in the previous cycle to be reflected in the state information of the second artificial intelligence model (210) in the current cycle, the second artificial intelligence model (210) can output touch information based on the touch data input in the current cycle and the previous state information. Therefore, the second artificial intelligence model (220) may be more suitable for outputting touch information for continuously input touch data than the first artificial intelligence model (210). For example, if a pocket touch is input three times in succession, the processor (140) can input touch data (303) corresponding to the third input pocket touch into the second artificial intelligence model (220) to output touch information (409) indicating an abnormal touch. Even if the pocket touch data (303) contains some features similar to the finger touch data, the second artificial intelligence model (220) of the current cycle can receive the previous state information generated by the second artificial intelligence model (220) in the previous cycle and output touch information based on the previous state information, and since the previous state information includes the features of the pocket touch data (301, 302) input in the previous cycle, the second artificial intelligence model (220) can output accurate touch information (409) that is an abnormal touch. However, when the type of touch data that is continuously input changes from pocket touch to finger touch, the second artificial intelligence model (220) may output inaccurate touch information. For example, if pocket touch data is inputted three times consecutively to the second artificial intelligence model (220) and then finger touch data (304) is inputted, the second artificial intelligence model (220) may output touch information (410) called an abnormal touch. The touch information (410) outputted by the second artificial intelligence model (220) in the current cycle is based on previous state information and currently input touch data, and since the previous state information includes features of pocket touch data (301, 302, 303), the second artificial intelligence model (220) may output incorrect touch information (410) called an abnormal touch. As described in the description of FIGS. 3 and 4, the first artificial intelligence model (210) is not suitable for outputting touch information for continuously input touch data, whereas the second artificial intelligence model (220) may be suitable for outputting touch information for continuously input touch data. In addition, the first artificial intelligence model (210) may output accurate touch information even when the type of continuously input touch data changes, whereas the second artificial intelligence model (220) may output inaccurate touch information when the type of continuously input touch data changes. The processor (140) can utilize the advantages of both artificial intelligence models by simultaneously inputting touch data to each of the first artificial intelligence model (210) and the second artificial intelligence model (220) and comparing the output touch information, thereby identifying accurate touch information and performing an operation corresponding to the touch information. The specific operation of the processor (140) comparing the touch information output by simultaneously inputting touch data to each of the first artificial intelligence model (210) and the second artificial intelligence model (220) will be described in detail in the description of FIG. 5. Meanwhile, the first artificial intelligence model (210) and the second artificial intelligence model (220) described above may be composed of a plurality of neural network layers. Each of the plurality of neural network layers has a plurality of weight values and performs neural network operations through operations between the operation results of the previous layer and the plurality of weights. The plurality of weights of the plurality of neural network layers may be optimized by the learning results of the first artificial intelligence model (210) and the second artificial intelligence model (220). For example, the plurality of weights may be updated so that the loss value or cost value obtained from the first artificial intelligence model (210) and the second artificial intelligence model (220) is reduced or minimized during the learning process. The artificial neural network may include a deep neural network (DNN), for example, a convolutional neural network (CNN), a deep neural network (DNN), a recurrent neural network (RNN), a restricted boltzmann machine (RBM), a deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), a long short term memory (LSTM), or deep Q-networks, but is not limited to the examples described above. FIG. 5 is a diagram illustrating a method for determining whether to initialize a second artificial intelligence model of an electronic device according to one or more embodiments of the present disclosure. As illustrated in FIG. 5, the processor (140) can input touch data (501, 502, 503, 504) to the first artificial intelligence model (210) and the second artificial intelligence model (220) to output touch information. The processor (140) can input three pocket touch data (501, 502, 503) sequentially into the first artificial intelligence model (210) to output touch information (509, 510, 511) called abnormal touch, and can input them into the second artificial intelligence model (220) to output touch information (505, 506, 507) called abnormal touch. In this case, since the touch information output through the first artificial intelligence model (210) and the touch information output through the second artificial intelligence model (220) are the same, the processor (140) can identify the touch information called abnormal touch as accurate touch information and perform an operation corresponding thereto. However, when the processor (140) inputs three pocket touch data (501, 502, 503) into the artificial intelligence model in succession, and then inputs finger touch data (504) into each of the first artificial intelligence model (210) and the second artificial intelligence model (220), the first artificial intelligence model (210) may output touch information (512) indicating a normal touch, and the second artificial intelligence model (220) may output touch information (508) indicating an abnormal touch. In this case, the processor (140) can identify that the touch information output by the first artificial intelligence model (210) and the second artificial intelligence model (220) is different, and can identify whether the amount of change in the sensing value of the sensor (130) exceeds a threshold value. According to one embodiment, let's assume a situation where the electronic device (100) is positioned inside a pocket and pocket touches are continuously input through the touch panel (120), and then the electronic device (100) moves outside the pocket and a finger touch is input to the touch panel (120). If the sensor (130) includes a light sensor and / or a motion detection sensor, when the electronic device (100) moves from the inside of the pocket to the outside, the amount of light incident on the light sensor may increase significantly, causing a significant change in the sensing value of the light sensor, or a significant change in the sensing value of the motion detection sensor may occur due to a rotational or linear movement of the electronic device (100). Accordingly, the processor (140) may identify that the electronic device (100) has moved from the inside of the pocket to the outside based on the fact that the amount of change in the sensing value exceeds a threshold value. Information about the threshold value may be set in advance and stored in the memory (110). At this time, the processor (140) can identify that a touch has been input by a finger from outside the pocket, and can identify the touch information (512) output by the first artificial intelligence model (210) as a normal touch as accurate touch information. In addition, the processor (140) can perform an operation based on the touch information (512) as a normal touch, and can initialize the state of the second artificial intelligence model (220). Here, "initializing the second artificial intelligence model (220)" may mean that the processor (140) controls the second artificial intelligence model (220) to output touch information based only on the touch data currently input. When the state of the second artificial intelligence model (220) is initialized, the second artificial intelligence model (220) may output touch information through an operation process that does not reflect previous state information. For example, if the previous state information is implemented in the form of a one-dimensional vector, the processor (140) can delete all previous state information by multiplying the previous state information by a zero vector to initialize the state of the second artificial intelligence model (220). However, this is only one example of a method for preventing the previous state information from affecting the output information of the second artificial intelligence model (220) of the next cycle, and the processor (140) can perform state initialization of the second artificial intelligence model (220) through various methods. According to one embodiment, the processor (140) can input touch data for a subsequent touch on the touch panel (120) into the initialized second artificial intelligence model (220) after the second artificial intelligence model (220) is initialized to identify whether the subsequent touch is a normal touch. FIG. 5 illustrates that pocket touch data (501, 502, 503) are input continuously and then finger touch data (504) is input, causing the second artificial intelligence model (220) to output incorrect touch information (508) indicating an abnormal touch. However, even when finger touch data is input continuously and then pocket touch data is input, the processor (140) can identify whether the amount of change in the sensing value of the sensor (130) exceeds a threshold value if the touch information output by the two artificial intelligence models is different. According to one embodiment, when the electronic device (100) is positioned outside the pocket, a user continuously inputs finger touches and then the electronic device (100) is moved inside the pocket. The processor (140) may continuously receive finger touch data and then input pocket touch data to the first artificial intelligence model (210) and the second artificial intelligence model (220). In this case, the first artificial intelligence model (210) may output accurate touch information called an abnormal touch for the pocket touch data, while the second artificial intelligence model (220) may output inaccurate touch information called a normal touch. At this time, since the touch information output by the two artificial intelligence models is different, the processor (140) can identify whether the amount of change in the sensing value of the sensor (130) exceeds a threshold value. For example, as the electronic device (100) moves from the outside to the inside of the pocket, the amount of light incident on the light sensor may decrease, and thus the sensing value may decrease. In this case, the amount of change in the sensing value of the light sensor may increase, or the amount of change in the sensing value of the motion detection sensor may increase. If the amount of change in the sensing value exceeds the threshold value, the processor (140) can identify that the electronic device (100) has moved from the outside to the inside of the pocket, and can identify the touch information called an abnormal touch output by the first artificial intelligence model (210) as accurate touch information, and perform an operation corresponding to the touch information called an abnormal touch. In addition, the state of the second artificial intelligence model (220) can be initialized by the above-described method. In addition, although FIG. 5 only illustrates that the touch information output by the second artificial intelligence model (220) may be inaccurate, the touch information output by the second artificial intelligence model (220) may be accurate and the touch information output by the first artificial intelligence model (210) may be inaccurate, and in this case, an action corresponding to the touch information output by the second artificial intelligence model (220) may be performed without initializing the state of the second artificial intelligence model (220). For example, if the finger touch data (504) of FIG. 5 is pocket touch data, the second artificial intelligence model (220) may output touch information indicating that all four touches are abnormal touches when four pocket touch data are consecutively input, while the first artificial intelligence model (210) may output touch information indicating that the fourth pocket touch data is a normal touch. In this case, since the touch information output by the two artificial intelligence models is different, the processor (140) may identify whether the amount of change in the sensing value of the sensor (130) exceeds the threshold value. At this time, since the electronic device (100) is only located inside the pocket and its location has not changed, the amount of change in the sensing value of the sensor (130), including a light sensor and / or a motion detection sensor, may also be below the threshold value. Since the change in the sensing value is below the threshold value, the processor (140) can identify the touch information output by the second artificial intelligence model (220) as accurate touch information and perform an operation corresponding to the touch information output by the second artificial intelligence model (220). In addition, the processor (140) may not perform state initialization of the second artificial intelligence model (220). In the above description, a method for determining whether to initialize the state of the second artificial intelligence model (220) based on the amount of change in the sensing value has been described, but the processor (140) can determine whether to initialize the state of the second artificial intelligence model (220) not only based on the amount of change in the sensing value, but also based on the probability value output by the second artificial intelligence model (220), which will be described in detail in the description of the drawings to be described later. FIG. 6 is a diagram illustrating a method for calculating an initialization score for initializing the state of a second artificial intelligence model of an electronic device according to one or more embodiments of the present disclosure. According to FIG. 6, the processor (140) can continuously input pocket touch data (601) and finger touch data (602) to the second artificial intelligence model (220) to output the probability that the input touch corresponds to an abnormal touch. According to one embodiment, the processor (140) can identify a difference between a first probability (P1) for touch information (603) output by the second artificial intelligence model (220) based on touch data (601) input in a previous cycle and a second probability (P2) for touch information (604) output based on touch data (602) input in a current cycle. Here, the "probability of touch information" may mean the probability that the touch data input to the second artificial intelligence model (220) corresponds to an abnormal touch. At this time, since the second artificial intelligence model (220) outputs the probability of whether the input touch data corresponds to an abnormal touch or a normal touch, the sum of the probability of corresponding to an abnormal touch and the probability of corresponding to a normal touch may always be 1, and the touch information may be output as corresponding to a touch with a larger probability value among them. For example, if the probability that the touch data input to the second artificial intelligence model (220) corresponds to a normal touch is output as 0.3, the probability of corresponding to an abnormal touch may be output as 0.7, and the second artificial intelligence model (220) may output touch information corresponding to an abnormal touch. Additionally, the “first probability” may mean the probability for touch information output by the second artificial intelligence model (220) in the previous cycle, and the “second probability” may mean the probability for touch information output by the second artificial intelligence model (220) in the current cycle. Therefore, even if the second artificial intelligence model (220) outputs the same touch information called an abnormal touch in the previous cycle and the current cycle, the first probability and the second probability may have different values. As illustrated in FIG. 6, if the second artificial intelligence model (220) outputs touch information (604) called an abnormal touch due to the influence of the pocket touch data (601) input in the previous cycle even though the finger touch data (602) was input to the second artificial intelligence model (220), the second probability may have a smaller value than the first probability. This is because the first probability is the probability that the second artificial intelligence model (220) predicted that the touch data (601) corresponds to an abnormal touch based on the pocket touch data (601), whereas the second probability is the probability that the second artificial intelligence model (220) predicted that the touch data (602) corresponds to an abnormal touch based on the finger touch data (602) and the previous state information. For example, if the first probability is 0.95, the second probability may be 0.67. Returning to Fig. 5, the touch information output by the first artificial intelligence model (210) and the touch information output by the second artificial intelligence model (220) may be different. In this case, if the change in the sensing value of the sensor (130) exceeds a threshold value, the processor (140) can identify the difference value (△P) between the first probability (P1) and the second probability (P2), and calculate a reset score based on the change in the sensing value and the difference value (△P) between the first probability and the second probability. As an example, the initialization score can be calculated by the following mathematical expression 1. Here, △S1 and △S2 may represent the amount of change in the sensing value of the first sensor and the amount of change in the sensing value of the second sensor, respectively. For example, if the sensor (130) includes a light sensor and a motion detection sensor, the first sensor may be a light sensor and the second sensor may be a motion detection sensor. At this time, △S1 and △S2 are normalized values to have values between 0 and 1. Here, △P can be the difference between the first probability and the second probability. Also, W1, W2 and W3 may represent weights assigned to △S1, △S2, and △P, respectively. W1, The weights of W2 and W3 can change depending on the situation and can have arbitrary values. For example, assume a situation where a user takes the electronic device (100) out of his / her pocket in a dark room and touches the touch panel (120) with his / her finger. In this case, since there was a rotational motion and a linear motion of the electronic device (100) moving from the inside of the pocket to the outside, the change amount (△S2) of the sensing value of the motion detection sensor, which is the second sensor, can indicate that the electronic device (100) has changed to a state where it has been taken out of the pocket. However, since the electronic device (100) continued to exist in the dark room, the change amount (△S1) of the sensing value of the light sensor, which is the first sensor, cannot indicate that the electronic device (100) has changed to a state where it has been taken out of the pocket. Therefore, in indicating a change in the state of the electronic device (100), since △S1 is a less important value than △S2, a small weight can be given to △S1 and a large weight can be given to △S2 (W1<< W2). In the above description, it has been exemplified that the value of the weight may change depending on the state of the electronic device (100), but this is only one example, and the processor (140) may adjust the size of the weight by identifying how many cycles have passed since the state initialization of the second artificial intelligence model (220) was performed, and may adjust the size of the weight in various ways. For example, if the state initialization of the second artificial intelligence model (220) was performed in the immediately previous cycle, a large weight may be assigned to △P (W1, W2 << W3), and if the state initialization of the second artificial intelligence model (220) was performed 200 or more cycles ago, a small weight may be assigned to △P (W3 << W1, W2). The processor (140) may initialize the state of the second artificial intelligence model (220) if the initialization score calculated by the above-described method exceeds a threshold score. For example, the processor (140) may initialize the state of the second artificial intelligence model (220) if the initialization score exceeds 0.5. Although the threshold score is set to 0.5 as an example, the threshold score may have any value between 0 and 1. According to another embodiment, if the initialization score has a value lower than or equal to a threshold score, the processor (140) may perform an operation corresponding to the touch information output by the first artificial intelligence model (210) without initializing the state of the second artificial intelligence model (220). FIG. 7 is a diagram illustrating a method for initializing a state of a second artificial intelligence model according to one or more embodiments of the present disclosure. According to FIG. 7, if no subsequent touch is input through the touch panel (120) for a preset period of time after touch information (703) is output by the second artificial intelligence model (220), the processor (140) can initialize the state of the second artificial intelligence model (220). According to one embodiment, the processor (140) may input pocket touch data (701) into the second artificial intelligence model (220) to obtain touch information (703) called an abnormal touch, and if subsequent touches input through the touch panel (120) are not identified for a period of 3 seconds or more thereafter, the processor may initialize the state of the second artificial intelligence model (220) by deleting previous state information or multiplying it by a zero value so as not to affect the operation process of the second artificial intelligence model (220) of the current cycle. In addition, the processor (140) can input subsequent touch data acquired after 3 seconds into the state-initialized second artificial intelligence model (220) to output touch information. In the above description, the preset time was described as 3 seconds, but this is only an example, and the preset time can be set to various times, such as 2 seconds, 4 seconds, etc., such as the time taken by the user to move the electronic device (100) from inside the pocket to outside the pocket, or the time taken to lift the electronic device (100) from the bottom surface of the desk to eye level. FIG. 8 is a diagram illustrating a data input method for a first artificial intelligence model and a second artificial intelligence model according to one or more embodiments of the present disclosure. According to FIG. 8, the processor (140) inputs touch data (800) acquired through the touch panel (120) into a convolutional neural network (201), and inputs the output value obtained into an LSTM neural network to output a probability (802) for touch information. Here, the first artificial intelligence model (210) and the second artificial intelligence model (220) may include both a convolutional neural network (201) and an LSTM neural network (202). Since the first artificial intelligence model (210) is modeled to output touch information that does not reflect previous state information in the current cycle, it may not include an LSTM neural network (202), whereas the second artificial intelligence model (220) outputs touch information by reflecting previous state information in the current cycle, and therefore must include a neural network having a cyclic structure, such as an LSTM neural network (202). According to one embodiment, the first artificial intelligence model (210) and the second artificial intelligence model (220) may be models having the same neural network structure including an LSTM neural network (202), but the first artificial intelligence model (210) may be trained to output touch information that does not reflect previous state information in the current cycle, while the second artificial intelligence model (220) may be trained to output touch information that reflects previous state information in the current cycle. According to another embodiment, the first artificial intelligence model (210) may be implemented as a model that does not include an LSTM neural network (202), and the second artificial intelligence model (220) may be implemented as a model that includes an LSTM neural network (202). In this case, the first artificial intelligence model (210) does not include an LSTM neural network (202) and thus cannot output touch information reflecting previous state information in the current cycle. On the other hand, the second artificial intelligence model (220) includes an LSTM neural network and thus can output touch information reflecting previous state information. Here, the touch data (800) may include information on at least one of the coordinates where the touch occurred and the amount of change in electrostatic capacity detected at the coordinates. For example, the touch data (800) may be implemented in the form of a two-dimensional array, and each element of the two-dimensional array may store the amount of change in electrostatic capacity detected by the sensing electrode present at each coordinate. In addition, the convolutional neural network (201) may include a convolution layer and a pooling layer. According to one embodiment, the processor (140) may input touch data (800) to the convolution layer, divide the touch data into a plurality of regions, and apply a filter that performs convolution on each of the divided regions. As a result of the convolution, the processor (140) may obtain feature values for the plurality of regions, and input a feature map composed of the feature values for the plurality of regions to the pooling layer. The feature map input to the pooling layer may reduce (downsample) the amount of information of the feature map input to the pooling layer through operations such as max pooling and average pooling. The processor (140) may obtain a feature map corresponding to the touch data (800) through the above-described convolutional neural network (201). In addition, the LSTM neural network (202) may include gates of an input gate, a forget gate, and an output gate. According to one embodiment, the processor (140) may input the feature map and previous state information obtained through the convolutional neural network (201) into the LSTM neural network (202). The processor (140) may generate temporary state information based on the input feature map and previous state information through the input gate. In addition, the processor (140) may output information on how much of the previous state information to remember, such as whether to completely forget or completely remember the input previous state information based on the input feature map through the forget gate. The processor (140) may generate current state information based on the temporary state information generated by the input gate, the information output from the forget gate, and the previous state information through the output gate, and output a probability (802) for touch information based on the current state information. As illustrated in FIG. 8, if the probability (802) information for touch information output by the artificial intelligence model indicates that the probability corresponding to a normal touch is 0.927 and the probability corresponding to an abnormal touch is 0.073, the processor (140) can identify that the input touch data (800) corresponds to a normal touch and perform an operation corresponding to a normal touch. The first artificial intelligence model (210) and the second artificial intelligence model (220) including a neural network structure as illustrated in FIG. 8 may be artificial intelligence models trained to identify whether a touch is normal based on at least one of a change in touch coordinates and a change in electrostatic capacitance input through the touch panel (120). For example, in the case of a normal touch inputted through a touch target such as a finger, stylus pen, or touch pen, the touch area may be small. In addition, since the user applies pressure to the touch panel (120) with the intention of touching, the touch intensity may be strong and the change in electrostatic capacity may be large accordingly. On the other hand, in the case where a touch is inputted to the touch panel (120) by a body part in contact with the pocket, or pressure is applied to the touch panel (120) by an object other than the touch target, the touch area may be larger than in the case of a normal touch, whereas the touch intensity may be weak due to the nature of an unintentional touch. Accordingly, an abnormal touch may have a larger touch area and a smaller change in electrostatic capacity than a normal touch. According to one embodiment, the first artificial intelligence model (210) and the second artificial intelligence model (220) can be trained to identify touch data having the characteristics of a small touch area and a large change in electrostatic capacitance as a normal touch, and to identify touch data having the characteristics of a large touch area and a small change in electrostatic capacitance as an abnormal touch. According to another embodiment, the first artificial intelligence model (210) and the second artificial intelligence model (220) may be trained to identify a touch data corresponding to a normal touch when the touch area is initially small and then widens over time, and to identify a touch data corresponding to an abnormal touch when the touch is made over a consistently wide area over time. Of course, the first artificial intelligence model (210) and the second artificial intelligence model (220) may be trained by various methods. FIG. 9 is a flowchart illustrating a method for controlling an electronic device according to one or more embodiments of the present disclosure. According to FIG. 9, in operation 910, the electronic device (100) can input touch data to the first artificial intelligence model and the second artificial intelligence model. Here, the first artificial intelligence model may refer to an artificial intelligence model trained to receive touch data input to the touch panel and output information on whether the input touch data corresponds to a normal touch or an abnormal touch. Unlike the first artificial intelligence model, the second artificial intelligence model may refer to an artificial intelligence model trained to output information on whether the touch data corresponds to a normal touch based on previous state information and touch data input in the current cycle. In operation 920, the electronic device (100) can identify whether the touch information output by the first artificial intelligence model and the second artificial intelligence model match. If the touch information output by the first artificial intelligence model and the second artificial intelligence model match (S920: Y), in operation 970, the electronic device (100) can perform an operation corresponding to the touch information output by the second artificial intelligence model. If the touch information output by the first artificial intelligence model and the second artificial intelligence model do not match (S920: N), in operation 930, the electronic device (100) can identify whether the amount of change in the sensing value of the sensor exceeds a threshold value. For example, the electronic device (100) can identify whether the amount of change in the sensing value of the sensor exceeds a threshold value when the touch information output by the first artificial intelligence model corresponds to a normal touch, while the touch information output by the second artificial intelligence model corresponds to an abnormal touch. Here, the sensor may include various sensors capable of detecting changes in the status of the electronic device, such as a light sensor and a motion detection sensor. If the amount of change in the sensing value does not exceed the threshold value (S930: N), in operation 970, the electronic device (100) can perform an operation corresponding to the touch information output by the second artificial intelligence model. If the amount of change in the sensing value exceeds the threshold value (S930: Y), in operation 940, the electronic device (100) can identify the difference value between the first probability and the second probability, and can calculate an initialization score based on the difference value between the first probability and the second probability and the amount of change in the sensing value. Next, at operation 950, the electronic device (100) can identify whether the initialization score exceeds a threshold score. If the initialization score does not exceed the threshold score (S950: N), in operation 970, the electronic device (100) can perform an operation corresponding to the touch information output by the second artificial intelligence model. For example, if the touch information output by the second artificial intelligence model corresponds to a normal touch, the electronic device (100) can perform various operations such as turning on or unlocking the power or executing / terminating an application. According to another example, if the touch information output by the second artificial intelligence model corresponds to an abnormal touch, the electronic device (100) can perform various actions, such as not turning on the power, displaying a malfunction prevention filter, or displaying a UI screen indicating that an abnormal touch is being input. If the initialization score exceeds the threshold score (S950: Y), in operation 960, the electronic device (100) can initialize the state of the second artificial intelligence model. Here, when the state initialization of the second artificial intelligence model is performed, the second artificial intelligence model can output touch information through an operation process that does not reflect the previous state information. For example, if the previous state information is implemented in the form of a one-dimensional vector, the electronic device (100) can delete all previous state information by multiplying the previous state information by a zero vector for the state initialization of the second artificial intelligence model. For example, when the state of the second artificial intelligence model is initialized, the electronic device (100) may input touch data for a subsequent touch on the touch panel into the initialized second artificial intelligence model to identify whether the subsequent touch is a normal touch. FIG. 10 is a flowchart illustrating a method for controlling an electronic device according to one or more embodiments of the present disclosure. According to FIG. 10, in operation 1010, when touch data is acquired, the electronic device (100) can input the touch data into a second artificial intelligence model to output touch information. In operation 1020, the electronic device (100) can identify that no subsequent touch has been input for a preset period of time after outputting touch information through the second artificial intelligence model. In operation 1030, if the electronic device (100) identifies that no subsequent touch has been input for a preset period of time, it may initialize the state of the second artificial intelligence model. In the description of FIGS. 1 to 10 described above, only an embodiment in which the electronic device (100) can identify whether a touch is normal by using the first artificial intelligence model and the second artificial intelligence model stored in the memory has been described, but the electronic device (100) may also perform the operations described in the description of FIGS. 1 to 10 described above by using various artificial intelligence models stored in the external server by linking with an external server. The various methods described in FIGS. 9 and 10 can be performed by an electronic device having the configuration shown in FIG. 2, but are not necessarily limited thereto, and can be performed by electronic devices having various configurations. Meanwhile, in FIGS. 9 and 10, the order is mapped for all steps for convenience of explanation, but it is of course not necessarily limited to the order of steps that are not related to the order or can be performed in parallel. The various embodiments of the present disclosure can be applied and implemented in all types of electronic devices, and each embodiment can be combined with each other in whole or in part to be applied to one device. Meanwhile, the various embodiments described above may be implemented in a computer-readable recording medium or similar device using software, hardware, or a combination thereof. In some cases, the embodiments described herein may be implemented by the processor itself. In a software implementation, embodiments, such as the procedures and functions described herein, may be implemented as separate software modules. Each of the software modules may perform one or more functions and operations described herein. Meanwhile, computer instructions for performing processing operations of the electronic device (100) according to various embodiments of the present disclosure described above may be stored in a non-transitory computer-readable medium. Computer instructions stored on such non-transitory computer-readable media, when executed by a processor of a specific device, cause the specific device to perform the setting methods according to the various embodiments described above. A non-transitory computer-readable medium refers to a medium that permanently stores data and can be read by a device, rather than a medium that stores data for a short period of time, such as a register, cache, or memory. Specific examples of non-transitory computer-readable media include CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, and ROMs. Although the preferred embodiments of the present disclosure have been illustrated and described above, the present disclosure is not limited to the specific embodiments described above, and various modifications may be made by a person having ordinary skill in the art to which the present disclosure pertains without departing from the gist of the present disclosure as claimed in the claims, and such modifications should not be understood individually from the technical idea or prospect of the present disclosure.

Claims

1. In electronic devices, A memory storing a first artificial intelligence model that outputs touch information based on touch data input in the current cycle and a second artificial intelligence model that outputs touch information based on the touch data input in the current cycle and previous state information; Touch panel; and comprising one or more processors; One or more of the above processors, Inputting touch data acquired through the touch panel into the first artificial intelligence model and the second artificial intelligence model, An electronic device that performs an operation corresponding to the touch information output from the first artificial intelligence model when the touch information output from the first artificial intelligence model and the second artificial intelligence model are different.

2. In paragraph 1, including sensors; One or more of the above processors, If the touch information output from the first artificial intelligence model and the second artificial intelligence model is different, the amount of change in the sensing value of the sensor during a predetermined unit time prior to the touch on the touch panel is identified, An electronic device that performs an action corresponding to the touch information output from the first artificial intelligence model when the above change amount exceeds a threshold value.

3. In paragraph 2 One or more of the above processors, An electronic device that performs an action corresponding to the touch information output from the second artificial intelligence model when the amount of change is less than or equal to the threshold value.

4. In paragraph 2 One or more of the above processors, If the above change amount exceeds the threshold value, the state of the second artificial intelligence model is initialized, An electronic device that inputs touch data for a subsequent touch on the touch panel into the initialized second artificial intelligence model to identify whether the subsequent touch is a normal touch.

5. In paragraph 2, One or more of the above processors, If the above change amount exceeds the above threshold, The second artificial intelligence model identifies the difference between the first probability for touch information output based on touch data acquired in the previous period and the second probability for touch information output based on touch data acquired in the current period, An electronic device that calculates a reset score based on the difference value between the first probability and the second probability and the change amount, and resets the state of the second artificial intelligence model when the reset score exceeds a threshold score.

6. In paragraph 2, The above sensor, At least one of a motion detection sensor and a light sensor, One or more of the above processors, Identify the sensing value of the sensor during the preset unit time prior to the touch on the touch panel and calculate the amount of change, An electronic device that initializes the state of the second artificial intelligence model when the amount of change exceeds the threshold value.

7. In paragraph 1, One or more of the above processors, An electronic device that initializes the state of the second artificial intelligence model when no subsequent touch is input to the touch panel for a predetermined time after touch information is output by the second artificial intelligence model.

8. In paragraph 1, The above first artificial intelligence model and the above second artificial intelligence model, At preset intervals, the touch data obtained through the touch panel is input and touch information on whether the touch is normal is output. The above previous status information is, Contains state information generated by the second artificial intelligence model in the previous period, The above current status information is: An electronic device comprising state information generated by the first artificial intelligence model or the second artificial intelligence model based on at least one of touch data acquired through the touch panel in the current period and the previous state information.

9. In paragraph 8, The above first artificial intelligence model and the above second artificial intelligence model, An electronic device comprising a convolutional neural network and a long short term memory (LSTM) neural network.

10. In paragraph 1, The above touch data is, Includes information on at least one of the coordinates at which a touch occurred on the touch panel and the amount of change in electrostatic capacity detected at the coordinates, The above first artificial intelligence model and the above second artificial intelligence model, An electronic device, which is an artificial intelligence model trained to identify whether a touch is normal based on at least one of a change in the above coordinates and a change in the above electrostatic capacitance.

11. In a method for controlling an electronic device, A step of inputting the acquired touch data into the first artificial intelligence model and the second artificial intelligence model; and A control method of an electronic device, comprising: a step of performing an operation corresponding to the touch information output from the first artificial intelligence model when the touch information output from the first artificial intelligence model and the second artificial intelligence model are different; 12. In paragraph 11, If the touch information output from the first artificial intelligence model and the second artificial intelligence model is different, a step of identifying the amount of change in the sensing value of the sensor during a predetermined unit time prior to a touch on the touch panel; and A control method of an electronic device, comprising: a step of performing an action corresponding to the touch information output from the first artificial intelligence model when the amount of change exceeds a threshold value.

13. In paragraph 12, A control method of an electronic device, comprising: a step of performing an action corresponding to the touch information output from the second artificial intelligence model when the amount of change is less than or equal to the threshold value.

14. In paragraph 12, If the change amount exceeds the threshold value, a step of initializing the state of the second artificial intelligence model; and A method for controlling an electronic device, comprising: inputting touch data for a subsequent touch on the touch panel into the initialized second artificial intelligence model to identify whether the subsequent touch is a normal touch; 15. A non-transitory computer-readable recording medium storing computer instructions that, when executed by a processor of an electronic device, cause the electronic device to perform an operation, the operation comprising: A step of inputting the acquired touch data into the first artificial intelligence model and the second artificial intelligence model; and A non-transitory computer-readable recording medium comprising: a step of performing an operation corresponding to the touch information output from the first artificial intelligence model when the touch information output from the first artificial intelligence model and the second artificial intelligence model are different;

Images