KR20220104595A - Method to reconfigure a machine learning cluster without interruption - Google Patents

Abstract

The present specification relates to a method for a first node to join a cluster for distributed learning, comprising: receiving a learning model for the distributed learning from a second node included in the cluster; enabling a background thread to participate in the cluster for the distributed learning, wherein the background thread generates a tensor related to an inclination value for the distributed learning; performing an Allreduce operation for the distributed learning through the tensor; and enabling first inclination values obtained through the Allreduce operation to be applied to the learning model.

Description

METHOD TO RECONFIGURE A MACHINE LEARNING CLUSTER WITHOUT INTERRUPTION

The present specification relates to a method and apparatus for reconstructing a machine learning cluster without interruption for efficiency of cloud resources in a distributed learning environment.

For synchronous distributed learning in machine learning, all computing nodes can train with the same learning model, and at the end of each iteration, the gradient values are collected and the average value is obtained to update the model.

Due to the nature of this learning process, existing distributed deep learning frameworks do not provide the ability to add computing nodes without interruption because it is difficult to synchronize the learning model without stopping learning.

An object of the present specification is to provide a method for adding a computing node in a distributed learning environment without interrupting the learning process while distributed learning is in progress.

In addition, an object of the present specification is to provide a method for synchronizing a learning model of a computing node to be newly added with a model in which learning is currently in progress in a distributed learning environment,

The technical problems to be achieved by the present specification are not limited to the technical problems mentioned above, and other technical problems not mentioned are clear to those of ordinary skill in the art to which this specification belongs from the detailed description of the invention below. will be able to be understood

An aspect of the present specification is a method in which a first node joins a cluster for distributed learning, the method comprising: receiving a learning model for distributed learning from a second node included in the cluster; joining a background thread to the cluster for distributed learning, wherein the background thread generates a tensor related to a gradient value for the distributed learning; performing an All-reduce operation for the distributed learning through the tensor; and applying the first gradient values obtained through the all-reduce operation to the learning model. may include.

Also, when the first node requests to join the cluster, the second node may stop the distributed learning in order to transfer the learning model to the first node.

Also, the tensor may have the same type as the gradient value for distributed learning, and may be initialized to 0.

In addition, the step of applying the first gradient values obtained through the all-reduce operation to the learning model may include: storing the obtained first gradient values in a memory; and sequentially applying the first gradient values stored in the memory to the learning model through a main thread. may further include.

In addition, receiving, from the nodes of the cluster, a second gradient value related to the next iterated distributed learning; applying the second gradient value to the learning model; and joining the cluster; may further include.

Another aspect of the present specification provides an apparatus for implementing a first node joining a cluster for distributed learning, comprising: a transceiver for transmitting and receiving a signal; Memory; and an AI processor for functionally controlling the transceiver and the memory. Including, wherein the AI processor receives the learning model for the distributed learning from the second node included in the cluster through the transceiver, and participates a background thread in the cluster for the distributed learning, , the background thread generates a tensor related to the gradient value for the distributed learning, and performs an All-reduce operation for the distributed learning through the tensor, and the all-reduce operation Through , the obtained first gradient values may be applied to the learning model.

According to an embodiment of the present specification, in a distributed learning environment, it is possible to provide a method for adding or removing a computing node without stopping the learning process during distributed learning.

In addition, it is an object of the present specification to provide a method of synchronizing a learning model of a computing node to be newly added with a model currently in progress in a distributed learning environment.

The effects obtainable in the present specification are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those of ordinary skill in the art to which this specification belongs from the description below. .

1 is a block diagram illustrating an electronic device related to the present specification.
2 is a block diagram of an AI device according to an embodiment of the present specification.
3 is an example of a distributed learning model to which this specification can be applied.
4 is an example of a cluster reconfiguration method to which the present specification can be applied.
5 is an embodiment to which the present specification can be applied.
6 is an example of a general apparatus to which this specification can be applied.
BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, which are included as part of the detailed description to help the understanding of the present specification, provide embodiments of the present specification, and together with the detailed description, explain the technical features of the present specification.

Hereinafter, the embodiments disclosed in the present specification will be described in detail with reference to the accompanying drawings, but the same or similar components are assigned the same reference numbers regardless of reference numerals, and redundant description thereof will be omitted. The suffixes "module" and "part" for components used in the following description are given or mixed in consideration of only the ease of writing the specification, and do not have distinct meanings or roles by themselves. In addition, in describing the embodiments disclosed in the present specification, if it is determined that detailed descriptions of related known technologies may obscure the gist of the embodiments disclosed in the present specification, the detailed description thereof will be omitted. In addition, the accompanying drawings are only for easy understanding of the embodiments disclosed in the present specification, and the technical spirit disclosed herein is not limited by the accompanying drawings, and all changes included in the spirit and scope of the present specification , should be understood to include equivalents or substitutes.

Terms including ordinal numbers such as first, second, etc. may be used to describe various elements, but the elements are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

When a component is referred to as being “connected” or “connected” to another component, it may be directly connected or connected to the other component, but it is understood that other components may exist in between. it should be On the other hand, when it is said that a certain element is "directly connected" or "directly connected" to another element, it should be understood that the other element does not exist in the middle.

The singular expression includes the plural expression unless the context clearly dictates otherwise.

In the present application, terms such as “comprises” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, but one or more other features It is to be understood that it does not preclude the possibility of the presence or addition of numbers, steps, operations, components, parts, or combinations thereof.

1 is a block diagram illustrating an electronic device related to the present specification.

Referring to FIG. 1 , the electronic device 100 includes a wireless communication unit 110 , an input unit 120 , a sensing unit 140 , an output unit 150 , an interface unit 160 , a memory 170 , and a control unit 180 . ) and a power supply 190 and the like. The components shown in FIG. 1 are not essential for implementing the electronic device, and thus the electronic device described herein may have more or fewer components than those listed above.

More specifically, the wireless communication unit 110 among the components, between the electronic device 100 and the wireless communication system, between the electronic device 100 and another electronic device 100, or the electronic device 100 and an external server It may include one or more modules that enable wireless communication between them. Also, the wireless communication unit 110 may include one or more modules for connecting the electronic device 100 to one or more networks.

The wireless communication unit 110 may include at least one of a broadcast reception module 111 , a mobile communication module 112 , a wireless Internet module 113 , a short-range communication module 114 , and a location information module 115 . .

The input unit 120 includes a camera 121 or an image input unit for inputting an image signal, a microphone 122 or an audio input unit for inputting an audio signal, and a user input unit 123 for receiving information from a user, for example, , a touch key, a push key, etc.). The voice data or image data collected by the input unit 120 may be analyzed and processed as a user's control command.

The sensing unit 140 may include one or more sensors for sensing at least one of information within the electronic device, surrounding environment information surrounding the electronic device, and user information. For example, the sensing unit 140 may include a proximity sensor 141, an illumination sensor 142, an illumination sensor, a touch sensor, an acceleration sensor, a magnetic sensor, and gravity. Sensor (G-sensor), gyroscope sensor, motion sensor, RGB sensor, infrared sensor (IR sensor: infrared sensor), fingerprint sensor (finger scan sensor), ultrasonic sensor , optical sensors (eg, cameras (see 121)), microphones (see 122), battery gauges, environmental sensors (eg, barometers, hygrometers, thermometers, radiation sensors, It may include at least one of a thermal sensor, a gas sensor, etc.) and a chemical sensor (eg, an electronic nose, a healthcare sensor, a biometric sensor, etc.). Meanwhile, the electronic device disclosed in the present specification may combine and utilize information sensed by at least two or more of these sensors.

The output unit 150 is for generating an output related to visual, auditory or tactile sense, and includes at least one of a display unit 151 , a sound output unit 152 , a haptip module 153 , and an optical output unit 154 . can do. The display unit 151 may implement a touch screen by forming a layer structure with the touch sensor or being formed integrally with the touch sensor. Such a touch screen may function as the user input unit 123 providing an input interface between the electronic device 100 and the user, and may provide an output interface between the electronic device 100 and the user.

The interface unit 160 serves as a passage with various types of external devices connected to the electronic device 100 . Such an interface unit 160, a wired / wireless headset port (port), an external charger port (port), a wired / wireless data port (port), a memory card (memory card) port, for connecting a device equipped with an identification module It may include at least one of a port, an audio I/O (Input/Output) port, a video I/O (Input/Output) port, and an earphone port. In response to the connection of the external device to the interface unit 160 , the electronic device 100 may perform appropriate control related to the connected external device.

In addition, the memory 170 stores data supporting various functions of the electronic device 100 . The memory 170 may store a plurality of application programs (or applications) driven in the electronic device 100 , data for operation of the electronic device 100 , and commands. At least some of these application programs may be downloaded from an external server through wireless communication. In addition, at least some of these application programs may exist on the electronic device 100 from the time of shipment for basic functions (eg, incoming calls, outgoing functions, message reception, and outgoing functions) of the electronic device 100 . Meanwhile, the application program may be stored in the memory 170 , installed on the electronic device 100 , and driven to perform an operation (or function) of the electronic device by the controller 180 .

In addition to the operation related to the application program, the controller 180 generally controls the overall operation of the electronic device 100 . The controller 180 may provide or process appropriate information or functions to the user by processing signals, data, information, etc. input or output through the above-described components or by driving an application program stored in the memory 170 .

Also, the controller 180 may control at least some of the components discussed with reference to FIG. 1 in order to drive an application program stored in the memory 170 . Furthermore, in order to drive the application program, the controller 180 may operate at least two or more of the components included in the electronic device 100 in combination with each other.

The power supply unit 190 receives external power and internal power under the control of the control unit 180 to supply power to each component included in the electronic device 100 . The power supply unit 190 includes a battery, and the battery may be a built-in battery or a replaceable battery.

At least some of the respective components may operate cooperatively to implement an operation, control, or control method of an electronic device according to various embodiments described below. In addition, the operation, control, or control method of the electronic device may be implemented on the electronic device by driving at least one application program stored in the memory 170 .

Hereinafter, before looking at various embodiments implemented through the electronic device 100 as described above, the above-listed components will be described in more detail.

First, referring to the wireless communication unit 110 , the broadcast reception module 111 of the wireless communication unit 110 receives a broadcast signal and/or broadcast related information from an external broadcast management server through a broadcast channel. The broadcast channel may include a satellite channel and a terrestrial channel. Two or more of the broadcast reception modules may be provided to the electronic device 100 for simultaneous broadcast reception or broadcast channel switching for at least two broadcast channels.

Mobile communication module 112, the technical standards or communication methods for mobile communication (eg, Global System for Mobile communication (GSM), Code Division Multi Access (CDMA), Code Division Multi Access 2000 (CDMA2000), EV -DO (Enhanced Voice-Data Optimized or Enhanced Voice-Data Only), WCDMA (Wideband CDMA), HSDPA (High Speed Downlink Packet Access), HSUPA (High Speed Uplink Packet Access), LTE (Long Term Evolution), LTE-A (Long Term Evolution-Advanced, etc.) transmits and receives radio signals to and from at least one of a base station, an external terminal, and a server on a mobile communication network constructed according to (Long Term Evolution-Advanced).

The wireless signal may include various types of data according to transmission/reception of a voice call signal, a video call signal, or a text/multimedia message.

The wireless Internet module 113 refers to a module for wireless Internet access, and may be built-in or external to the electronic device 100 . The wireless Internet module 113 is configured to transmit and receive wireless signals in a communication network according to wireless Internet technologies.

As wireless Internet technology, for example, WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), High Speed Downlink Packet Access (HSDPA), High Speed Uplink Packet Access (HSUPA), Long Term Evolution (LTE), Long Term Evolution-Advanced (LTE-A), etc., and the wireless Internet module ( 113) transmits and receives data according to at least one wireless Internet technology within a range including Internet technologies not listed above.

From the point of view that wireless Internet access by WiBro, HSDPA, HSUPA, GSM, CDMA, WCDMA, LTE, LTE-A, etc. is made through a mobile communication network, the wireless Internet module 113 performs wireless Internet access through the mobile communication network. ) may be understood as a type of the mobile communication module 112 .

Short-range communication module 114 is for short-range communication, Bluetooth (Bluetooth??), RFID (Radio Frequency Identification), infrared communication (Infrared Data Association; IrDA), UWB (Ultra Wideband), ZigBee, At least one of Near Field Communication (NFC), Wireless-Fidelity (Wi-Fi), Wi-Fi Direct, Wireless Universal Serial Bus (USB), and Magnetic Secure Transmission (MST) may be used to support short-distance communication. . The short-distance communication module 114, between the electronic device 100 and the wireless communication system, between the electronic device 100 and another electronic device 100, or the electronic device 100 through wireless area networks (Wireless Area Networks) ) and a network in which another electronic device 100 or an external server is located may support wireless communication. The local area network may be local area networks (Wireless Personal Area Networks).

Here, the other electronic device 100 is a wearable device capable of exchanging (or interworking) data with the electronic device 100 according to the present specification, for example, a smart watch, smart glasses. (smart glass), HMD (head mounted display)). The short-range communication module 114 may detect (or recognize) a wearable device capable of communicating with the electronic device 100 in the vicinity of the electronic device 100 . Furthermore, when the sensed wearable device is a device authenticated to communicate with the electronic device 100 according to the present specification, the controller 180 transmits at least a portion of data processed by the electronic device 100 to the short-range communication module ( 114) to transmit to the wearable device. Accordingly, the user of the wearable device may use data processed by the electronic device 100 through the wearable device. For example, according to this, when a call is received in the electronic device 100, the user performs a phone call through the wearable device, or when a message is received in the electronic device 100, the user receives the received call through the wearable device. It is possible to check the message.

The location information module 115 is a module for acquiring a location (or current location) of an electronic device, and a representative example thereof includes a Global Positioning System (GPS) module or a Wireless Fidelity (WiFi) module. For example, if the electronic device utilizes a GPS module, the electronic device may acquire the location of the electronic device by using a signal transmitted from a GPS satellite. As another example, if the electronic device utilizes the Wi-Fi module, it may acquire the location of the electronic device based on information of the Wi-Fi module and a wireless access point (AP) that transmits or receives a wireless signal. If necessary, the location information module 115 may perform any function of the other modules of the wireless communication unit 110 to obtain data on the location of the electronic device as a substitute or additionally. The location information module 115 is a module used to obtain the location (or current location) of the electronic device, and is not limited to a module that directly calculates or obtains the location of the electronic device.

Next, the input unit 120 is for input of image information (or signal), audio information (or signal), data, or information input from a user. For input of image information, the electronic device 100 has one Alternatively, a plurality of cameras 121 may be provided. The camera 121 processes an image frame such as a still image or a moving image obtained by an image sensor in a video call mode or a shooting mode. The processed image frame may be displayed on the display unit 151 or stored in the memory 170 . On the other hand, the plurality of cameras 121 provided in the electronic device 100 may be arranged to form a matrix structure, and through the cameras 121 forming the matrix structure in this way, various angles or focal points are applied to the electronic device 100 . A plurality of image information having a plurality of images may be input. In addition, the plurality of cameras 121 may be arranged in a stereo structure to acquire a left image and a right image for realizing a stereoscopic image.

The microphone 122 processes an external sound signal as electrical voice data. The processed voice data may be variously utilized according to a function (or a running application program) being performed by the electronic device 100 . Meanwhile, various noise removal algorithms for removing noise generated in the process of receiving an external sound signal may be implemented in the microphone 122 .

The user input unit 123 is for receiving information from a user, and when information is input through the user input unit 123 , the controller 180 may control the operation of the electronic device 100 to correspond to the input information. . Such, the user input unit 123 is a mechanical input means (or a mechanical key, for example, a button located on the front, rear or side of the electronic device 100, a dome switch (dome switch), a jog wheel, jog switch, etc.) and a touch-type input means. As an example, the touch input means consists of a virtual key, a soft key, or a visual key displayed on the touch screen through software processing, or a part other than the touch screen. It may be made of a touch key (touch key) disposed on the. On the other hand, the virtual key or the visual key, it is possible to be displayed on the touch screen while having various forms, for example, a graphic (graphic), text (text), an icon (icon), a video (video) or these can be made by a combination of

Meanwhile, the sensing unit 140 senses at least one of information in the electronic device, surrounding environment information surrounding the electronic device, and user information, and generates a sensing signal corresponding thereto. The controller 180 may control the driving or operation of the electronic device 100 or perform data processing, function, or operation related to an application program installed in the electronic device 100 based on the sensing signal. Representative sensors among various sensors that may be included in the sensing unit 140 will be described in more detail.

First, the proximity sensor 141 refers to a sensor that detects the presence or absence of an object approaching a predetermined detection surface or an object existing in the vicinity without mechanical contact using the force of an electromagnetic field or infrared rays. The proximity sensor 141 may be disposed in an inner region of the electronic device covered by the touch screen as described above or near the touch screen.

Examples of the proximity sensor 141 include a transmission type photoelectric sensor, a direct reflection type photoelectric sensor, a mirror reflection type photoelectric sensor, a high frequency oscillation type proximity sensor, a capacitive type proximity sensor, a magnetic type proximity sensor, an infrared proximity sensor, and the like. When the touch screen is capacitive, the proximity sensor 141 may be configured to detect the proximity of an object having conductivity by a change in an electric field according to the proximity of the object. In this case, the touch screen (or touch sensor) itself may be classified as a proximity sensor.

On the other hand, for convenience of description, the act of approaching an object on the touch screen without contacting it so that the object is recognized that it is located on the touch screen is called “proximity touch”, and the touch The act of actually touching an object on the screen is called "contact touch". The position where the object is touched in proximity on the touch screen means a position where the object is perpendicular to the touch screen when the object is touched in proximity. The proximity sensor 141 may detect a proximity touch and a proximity touch pattern (eg, proximity touch distance, proximity touch direction, proximity touch speed, proximity touch time, proximity touch position, proximity touch movement state, etc.) have. Meanwhile, the controller 180 processes data (or information) corresponding to the proximity touch operation and the proximity touch pattern sensed through the proximity sensor 141 as described above, and furthermore, provides visual information corresponding to the processed data. It can be printed on the touch screen. Furthermore, the controller 180 may control the electronic device 100 to process different operations or data (or information) according to whether a touch to the same point on the touch screen is a proximity touch or a contact touch. .

The touch sensor receives a touch (or touch input) applied to the touch screen (or the display unit 151) using at least one of various touch methods such as a resistive film method, a capacitive method, an infrared method, an ultrasonic method, and a magnetic field method. detect

As an example, the touch sensor may be configured to convert a change in pressure applied to a specific part of the touch screen or a change in capacitance occurring in a specific part of the touch screen into an electrical input signal. The touch sensor may be configured to detect a position, an area, a pressure at the time of touch, capacitance at the time of touch, and the like where a touch object applying a touch on the touch screen is touched on the touch sensor. Here, the touch object is an object that applies a touch to the touch sensor, and may be, for example, a finger, a touch pen, a stylus pen, or a pointer.

As such, when there is a touch input to the touch sensor, a signal(s) corresponding thereto is sent to the touch controller. The touch controller processes the signal(s) and then transmits corresponding data to the controller 180 . Accordingly, the controller 180 can know which area of the display unit 151 has been touched, and the like. Here, the touch controller may be a component separate from the controller 180 , or may be the controller 180 itself.

Meanwhile, the controller 180 may perform different controls or may perform the same control according to the type of the touch object that touches the touch screen (or a touch key provided in addition to the touch screen). Whether to perform different control or the same control according to the type of the touch object may be determined according to the current operating state of the electronic device 100 or a running application program.

On the other hand, the above-described touch sensor and proximity sensor independently or in combination, a short (or tap) touch on the touch screen (short touch), long touch (long touch), multi touch (multi touch), drag touch (drag touch) ), flick touch, pinch-in touch, pinch-out touch, swype touch, hovering touch, etc. It can sense touch.

The ultrasonic sensor may recognize location information of a sensing target by using ultrasonic waves. Meanwhile, the controller 180 may calculate the position of the wave source based on information sensed by the optical sensor and the plurality of ultrasonic sensors. The position of the wave source may be calculated using the property that light is much faster than ultrasonic waves, that is, the time at which light arrives at the optical sensor is much faster than the time at which ultrasonic waves reach the ultrasonic sensor. More specifically, the position of the wave source may be calculated by using a time difference from the time that the ultrasonic wave arrives using light as a reference signal.

On the other hand, the camera 121 as seen in the configuration of the input unit 120 includes at least one of a camera sensor (eg, CCD, CMOS, etc.), a photo sensor (or an image sensor), and a laser sensor.

The camera 121 and the laser sensor may be combined with each other to detect a touch of a sensing target for a 3D stereoscopic image. The photo sensor may be stacked on the display device, and the photo sensor is configured to scan the motion of the sensing target close to the touch screen. More specifically, the photo sensor mounts photo diodes and transistors (TRs) in rows/columns and scans the contents placed on the photo sensor using electrical signals that change according to the amount of light applied to the photo diodes. That is, the photo sensor calculates the coordinates of the sensing target according to the amount of change in light, and through this, location information of the sensing target can be obtained.

The display unit 151 displays (outputs) information processed by the electronic device 100 . For example, the display unit 151 may display execution screen information of an application program driven in the electronic device 100 or UI (User Interface) and GUI (Graphic User Interface) information according to the execution screen information. .

Also, the display unit 151 may be configured as a stereoscopic display unit for displaying a stereoscopic image.

A three-dimensional display method such as a stereoscopic method (glasses method), an auto stereoscopic method (glassesless method), or a projection method (holographic method) may be applied to the stereoscopic display unit.

The sound output unit 152 may output audio data received from the wireless communication unit 110 or stored in the memory 170 in a call signal reception, a call mode or a recording mode, a voice recognition mode, a broadcast reception mode, and the like. The sound output unit 152 also outputs a sound signal related to a function (eg, a call signal reception sound, a message reception sound, etc.) performed by the electronic device 100 . The sound output unit 152 may include a receiver, a speaker, a buzzer, and the like.

The haptic module 153 generates various tactile effects that the user can feel. A representative example of the tactile effect generated by the haptic module 153 may be vibration. The intensity and pattern of vibration generated by the haptic module 153 may be controlled by a user's selection or setting of the controller. For example, the haptic module 153 may synthesize and output different vibrations or output them sequentially.

In addition to vibration, the haptic module 153 is configured to respond to stimuli such as a pin arrangement that moves vertically with respect to the contact skin surface, a jet or suction force of air through a nozzle or an inlet, a brush against the skin surface, contact of an electrode, an electrostatic force, etc. Various tactile effects can be generated, such as an effect caused by heat absorption and an effect by reproducing a feeling of cold and heat using an element capable of absorbing heat or generating heat.

The haptic module 153 may not only deliver a tactile effect through direct contact, but may also be implemented so that the user can feel the tactile effect through a muscle sense such as a finger or arm. Two or more haptic modules 153 may be provided according to the configuration of the electronic device 100 .

The light output unit 154 outputs a signal for notifying the occurrence of an event by using the light of the light source of the electronic device 100 . Examples of the event generated in the electronic device 100 may be message reception, call signal reception, missed call, alarm, schedule notification, email reception, information reception through an application, and the like.

The signal output from the light output unit 154 is realized when the electronic device emits light of a single color or a plurality of colors toward the front or rear. The signal output may be terminated when the electronic device detects the user's event confirmation.

The interface unit 160 serves as a passage with all external devices connected to the electronic device 100 . The interface unit 160 receives data from an external device, receives power and transmits it to each component inside the electronic device 100 , or transmits data inside the electronic device 100 to an external device. For example, a wired/wireless headset port, an external charger port, a wired/wireless data port, a memory card port, a port for connecting a device equipped with an identification module (port), audio I/O (Input/Output) port (port), video I/O (Input/Output) port (port), earphone port (port), etc. may be included in the interface unit 160 .

On the other hand, the identification module is a chip storing various information for authenticating the use authority of the electronic device 100, a user identification module (UIM), a subscriber identity module (subscriber identity module; SIM), universal user authentication It may include a universal subscriber identity module (USIM) and the like. A device equipped with an identification module (hereinafter, 'identification device') may be manufactured in the form of a smart card. Accordingly, the identification device may be connected to the terminal 100 through the interface unit 160 .

In addition, when the electronic device 100 is connected to an external cradle, the interface unit 160 serves as a passage through which power from the cradle is supplied to the electronic device 100 or is input from the cradle by a user. It may be a path through which various command signals are transmitted to the electronic device 100 . The various command signals or the power input from the cradle may be operated as signals for recognizing that the electronic device 100 is correctly mounted on the cradle.

The memory 170 may store a program for the operation of the controller 180 and may temporarily store input/output data (eg, a phone book, a message, a still image, a moving picture, etc.). The memory 170 may store data related to vibrations and sounds of various patterns output when a touch input is performed on the touch screen.

Memory 170 is a flash memory type (flash memory type), a hard disk type (hard disk type), an SSD type (Solid State Disk type), an SDD type (Silicon Disk Drive type), a multimedia card micro type ), card-type memory (such as SD or XD memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read (EEPROM) -only memory), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk may include at least one type of storage medium. The electronic device 100 may be operated in relation to a web storage that performs a storage function of the memory 170 on the Internet.

Meanwhile, as described above, the controller 180 controls an operation related to an application program and generally an overall operation of the electronic device 100 . For example, when the state of the electronic device satisfies a set condition, the controller 180 may execute or release a lock state that restricts input of a user's control command to applications.

In addition, the controller 180 performs control and processing related to voice calls, data communication, video calls, etc., or performs pattern recognition processing capable of recognizing handwriting input or drawing input performed on the touch screen as characters and images, respectively. can Furthermore, in order to implement various embodiments described below on the electronic device 100 according to the present specification, the controller 180 may control any one or a plurality of components described above by combining them.

The power supply unit 190 receives external power and internal power under the control of the control unit 180 to supply power required for operation of each component. The power supply unit 190 includes a battery, and the battery may be a built-in battery configured to be rechargeable, and may be detachably coupled to the terminal body for charging or the like.

In addition, the power supply unit 190 may include a connection port, and the connection port may be configured as an example of the interface 160 to which an external charger that supplies power for charging the battery is electrically connected.

As another example, the power supply unit 190 may be configured to charge the battery in a wireless manner without using the connection port. In this case, the power supply unit 190 using one or more of an inductive coupling method based on a magnetic induction phenomenon or a resonance coupling method based on an electromagnetic resonance phenomenon from an external wireless power transmitter. power can be transmitted. In this specification, the electronic device 100 may include a terminal and a server.

2 is a block diagram of an AI device according to an embodiment of the present specification.

The AI device 20 may include an electronic device including an AI module capable of performing AI processing, or a server including the AI module. Also, the AI device 20 may be included as a component of at least a part of the electronic device 100 shown in FIG. 1 to perform at least a part of AI processing together.

The AI device 20 may include an AI processor 21 , a memory 25 and/or a communication unit 27 .

The AI device 20 is a computing device capable of learning a neural network, and may be implemented in various electronic devices such as a server, a desktop PC, a notebook PC, and a tablet PC.

The AI processor 21 may learn the neural network using a program stored in the memory 25 . In particular, the AI processor 21 may learn a neural network for recognizing vehicle-related data. Here, the neural network for recognizing vehicle-related data may be designed to simulate a human brain structure on a computer, and may include a plurality of network nodes having weights that simulate neurons of the human neural network. The plurality of network modes may transmit and receive data according to a connection relationship, respectively, so as to simulate a synaptic activity of a neuron in which a neuron sends and receives a signal through a synapse. Here, the neural network may include a deep learning model developed from a neural network model. In a deep learning model, a plurality of network nodes can exchange data according to a convolutional connection relationship while being located in different layers. Examples of neural network models include deep neural networks (DNN), convolutional deep neural networks (CNN), Recurrent Boltzmann Machine (RNN), Restricted Boltzmann Machine (RBM), deep trust It includes various deep learning techniques such as neural networks (DBN, deep belief networks) and deep Q-networks, and can be applied to fields such as computer vision, speech recognition, natural language processing, and voice/signal processing.

Meanwhile, the processor performing the above-described function may be a general-purpose processor (eg, CPU), but may be an AI-only processor (eg, GPU) for artificial intelligence learning.

The memory 25 may store various programs and data necessary for the operation of the AI device 20 . The memory 25 may be implemented as a non-volatile memory, a volatile memory, a flash-memory, a hard disk drive (HDD), or a solid state drive (SDD). The memory 25 is accessed by the AI processor 21 , and reading/writing/modification/deletion/update of data by the AI processor 21 may be performed. Also, the memory 25 may store a neural network model (eg, the deep learning model 26 ) generated through a learning algorithm for data classification/recognition according to an embodiment of the present specification.

Meanwhile, the AI processor 21 may include a data learning unit 22 that learns a neural network for data classification/recognition. The data learning unit 22 may learn a criterion regarding which training data to use to determine data classification/recognition and how to classify and recognize data using the training data. The data learning unit 22 may learn the deep learning model by acquiring learning data to be used for learning and applying the acquired learning data to the deep learning model.

The data learning unit 22 may be manufactured in the form of at least one hardware chip and mounted on the AI device 20 . For example, the data learning unit 22 may be manufactured in the form of a dedicated hardware chip for artificial intelligence (AI), or may be manufactured as a part of a general-purpose processor (CPU) or graphics-only processor (GPU) to the AI device 20 . may be mounted. In addition, the data learning unit 22 may be implemented as a software module. When implemented as a software module (or a program module including instructions), the software module may be stored in a computer-readable non-transitory computer readable medium. In this case, the at least one software module may be provided by an operating system (OS) or may be provided by an application.

The data learning unit 22 may include a training data acquiring unit 23 and a model learning unit 24 .

The training data acquisition unit 23 may acquire training data required for a neural network model for classifying and recognizing data. For example, the training data acquisition unit 23 may acquire vehicle data and/or sample data to be input to the neural network model as training data.

The model learning unit 24 may use the acquired training data to learn the neural network model to have a criterion for determining how to classify predetermined data. In this case, the model learning unit 24 may train the neural network model through supervised learning using at least a portion of the training data as a criterion for determination. Alternatively, the model learning unit 24 may learn the neural network model through unsupervised learning for discovering a judgment criterion by self-learning using learning data without guidance. Also, the model learning unit 24 may train the neural network model through reinforcement learning using feedback on whether the result of the situation determination according to the learning is correct. Also, the model learning unit 24 may train the neural network model by using a learning algorithm including an error back-propagation method or a gradient decent method.

When the neural network model is trained, the model learning unit 24 may store the learned neural network model in a memory. The model learning unit 24 may store the learned neural network model in the memory of the server connected to the AI device 20 through a wired or wireless network.

The data learning unit 22 further includes a training data preprocessing unit (not shown) and a training data selection unit (not shown) in order to improve the analysis result of the recognition model or to save resources or time required for generating the recognition model. You may.

The learning data preprocessor may preprocess the acquired data so that the acquired data can be used for learning for situation determination. For example, the training data preprocessor may process the acquired data into a preset format so that the model learning unit 24 may use the acquired training data for image recognition learning.

In addition, the learning data selection unit may select data necessary for learning from among the learning data acquired by the learning data acquiring unit 23 or the training data pre-processed by the preprocessing unit. The selected training data may be provided to the model learning unit 24 . For example, the learning data selector may select only data about an object included in the specific region as the learning data by detecting a specific region among images acquired through a vehicle camera.

In addition, the data learning unit 22 may further include a model evaluation unit (not shown) in order to improve the analysis result of the neural network model.

The model evaluator may input evaluation data to the neural network model and, when an analysis result output from the evaluation data does not satisfy a predetermined criterion, may cause the model learning unit 22 to learn again. In this case, the evaluation data may be predefined data for evaluating the recognition model. As an example, the model evaluation unit may evaluate as not satisfying a predetermined criterion when, among the analysis results of the learned recognition model for the evaluation data, the number or ratio of evaluation data for which the analysis result is not accurate exceeds a preset threshold value. have.

The communication unit 27 may transmit the AI processing result by the AI processor 21 to an external electronic device.

Here, the external electronic device may be defined as a terminal or a client. In addition, the AI device 20 may be implemented through a server or a network.

On the other hand, although the AI device 20 shown in FIG. 2 has been functionally divided into the AI processor 21, the memory 25, the communication unit 27, and the like, the above-described components are integrated into one module and the AI module Note that it may also be called

3 is an example of a distributed learning model to which this specification can be applied.

Referring to FIG. 3 , nodes for distributed learning form a ring structure, and are connected through a framework, such as Tensorflow or Pytorch, to perform distributed learning. In order to collect the gradient values of nodes connected in such a ring structure, the Ring Allreduce method may be used. Ring Allreduce is an operation that reduces the target array of all processors to a single array and returns the resulting array to all processes.

For example, the collected gradient value may be divided by the number of nodes. This divided set of gradient values may be referred to as a chunk. Each node can pass one chunk to the next node in parallel, and the delivered chunk can be combined with each node's chunk.

That is, each node can add the local chunk to the received chunk and forward it to the next node. All chunks are passed through the ring, and chunks can be merged at each node. Through this, when a chunk visits all nodes (eg, when the process is performed as many as the number of nodes), each node may have one chunk in which the results of all nodes are combined.

A method in which computing nodes are added or removed in such a distributed learning model may be as follows.

1. Checkpoint method: Basically, when the node configuration is changed, learning is stopped and learning is resumed with the changed configuration. At this time, the computing node that has not previously participated in training does not have an up-to-date training model, so the training model must be saved (checkpointed) in the form of a file when training is stopped and delivered to the new computing node. .

2. Broadcast method: When there is a demand to change the node configuration during training, one of the existing computing nodes transmits information including the changed configuration and learning model to all nodes using broadcast communication. After completing this synchronization process, learning resumes.

In method 1, learning is stopped from the time the learning model is saved until learning is resumed with a new configuration, and in method 2, learning is stopped until the synchronization process of delivering the learning model through broadcast communication is finished. Therefore, it is not possible to fully utilize the changing computing resources used in the long-term learning process. That is, the efficiency of providing cloud resources is greatly reduced. In particular, since the section where the largest processing time occurs is the part where the learning model is delivered to a new computing node, a new technique to minimize this time is needed.

4 is an example of a cluster reconfiguration method to which the present specification can be applied.

Referring to FIG. 4 , the cluster reconfiguration method of the present specification may be performed through a background thread and a main thread of a node to be added.

1. When a new node is added while learning of existing clusters is in progress, the learning model synchronization process

For example, when a new node addition is requested to the existing clusters, one of the existing computing nodes may finish the ongoing learning and transfer the model learned so far to the additional requested node. At this time, since the learned model should not be changed during the transfer process, the node that transfers the learned model may be temporarily excluded from the learning process.

The node to be added participates in the communication pool to which the existing computing nodes are connected to receive the gradient values calculated later while receiving the learning model. In addition, the background thread of the node that delivers the learned model can also participate in the communication pool to which the existing computing nodes are connected, and the operation of the node to be added next can be performed in the same way. This is because, after the node to be added receives the learning model, the learning model changes as the learning of the existing computing nodes proceeds while performing the learning preparation process and loading the learning model.

The background thread of the node to be added prepares a tensor of the same type as the gradient value exchanged by existing computing nodes, and initializes the value to 0. When the existing computing nodes perform the Allreduce operation to obtain the average value of the gradient, the Background Thread can participate in the Allreduce operation through the previously prepared tensor. That is, since the Allreduce operation is derived as a result of the sum of the tensors transmitted by each node, the node to be added can receive the same gradient value as the existing computing nodes without affecting the learning operation process.

The node to be added may sequentially store the gradient values secured through the above process in the memory, and the Main Thread may secure the latest learning model by sequentially applying the gradient values stored in the memory to the learning model.

2. The process of joining the cluster of nodes with the latest learning model

The node to be added that has secured the latest learning model can apply the last received gradient value to the latest learning model as soon as the existing nodes finish the next iteration learning. Upon application, the new node participates in learning along with the existing computing nodes. After that, the new node can perform Allreduce operation with the gradient value calculated through training without preparing the tensor to receive the gradient value.

Through this, in the present specification, the configuration of the learning cluster can be freely changed during the learning process, so that it is possible to respond to changes in computing resources used in the learning process performed for a long time, and the efficiency of cloud resource use can be greatly improved.

5 is an embodiment to which the present specification can be applied.

Referring to FIG. 5 , the node may be implemented physically or virtualized by the electronic device 100 . The electronic device 100 may include a graphic processing unit (GPU).

The first node to be added receives a learning model for distributed learning from the second node included in the existing cluster (S510). For example, when the first node requests to join the cluster, the second node may stop the distributed learning in order to transfer the learning model to the first node. This can be managed by the system node that manages the distributed learning.

The first node joins the background thread to the cluster for distributed learning (S520). In more detail, the background thread may generate a tensor having the same shape as the gradient value used for distributed learning. For example, such a tensor has the same data type as the gradient value for distributed learning, and each data field may be initialized to 0.

The first node performs an All-reduce operation for distributed learning through the tensor (S530).

The first node applies the obtained gradient value to the learning model through an all-reduce operation (S540). For example, the first node may store the gradient values obtained in the all-reduce operation in the memory through the tensor, and apply them to the learning model in order in the main thread. Through this, the first node can acquire the latest learning model without interrupting the learning process of the existing cluster.

Thereafter, the first node receives the distributed learning gradient value through the next iteration and applies it to the learning model obtained immediately before. Thereafter, the first node joins the cluster and can perform distributed learning like the nodes included in the existing cluster without generating a separate tensor.

General devices to which this specification may be applied

Referring to FIG. 6 , the server X200 according to the proposed embodiment may include a communication module X210 , a processor X220 , and a memory X230 . The communication module X210 is also referred to as a radio frequency (RF) unit. The communication module X210 may be configured to transmit various signals, data and information to an external device and to receive various signals, data and information to an external device. The server X200 may be connected to an external device by wire and/or wirelessly. The communication module X210 may be implemented by being separated into a transmitter and a receiver. The processor X220 may control the overall operation of the server X200, and the server X200 may be configured to perform a function of calculating and processing information to be transmitted and received with an external device. In addition, the processor X220 may be configured to perform the server operation proposed in this specification. The processor X220 may control the communication module X110 to transmit data or a message to the UE, another vehicle, or another server according to the proposal of the present specification. The memory X230 may store arithmetic-processed information and the like for a predetermined time, and may be replaced with a component such as a buffer.

In addition, the terminal device X100 and the server X200 as described above may include a GPU, and their specific configuration, as described in the various embodiments of the present specification described above, are independently applied or two or more embodiments are simultaneously applied. It may be implemented to be applied, and the overlapping content will be omitted for clarity.

The above-described specification can be implemented as computer-readable code on a medium in which a program is recorded. The computer-readable medium includes all kinds of recording devices in which data readable by a computer system is stored. Examples of computer-readable media include Hard Disk Drive (HDD), Solid State Disk (SSD), Silicon Disk Drive (SDD), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. There is also a carrier wave (eg, transmission over the Internet) that includes implementation in the form of. Accordingly, the above detailed description should not be construed as restrictive in all respects but as exemplary. The scope of this specification should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of this specification are included in the scope of this specification.

In addition, although the above description has been focused on services and embodiments, this is only an example and does not limit the present specification, and those of ordinary skill in the art to which this specification pertains within a range that does not deviate from the essential characteristics of the present service and embodiments. It can be seen that various modifications and applications not exemplified above are possible. For example, each component specifically shown in the embodiments may be implemented by modification. And differences related to such modifications and applications should be construed as being included in the scope of the present specification defined in the appended claims.

Claims (10)

In the method for the first node to join a cluster for distributed learning,
receiving a learning model for the distributed learning from a second node included in the cluster;
joining a background thread to the cluster for distributed learning, wherein the background thread generates a tensor related to a gradient value for the distributed learning;
performing an All-reduce operation for the distributed learning through the tensor; and
applying first gradient values obtained through the all-reduce operation to the learning model;
A method comprising
According to claim 1,
the second node
When the first node requests to join the cluster, stopping the distributed learning to deliver the learning model to the first node.
According to claim 1,
The tensor has the same type as the gradient value for the distributed learning, and is initialized to 0, the method.
According to claim 1,
The step of applying the first gradient values obtained through the all-reduce operation to the learning model includes:
storing the obtained first gradient values in a memory; and
applying the first gradient values stored in the memory to the learning model in order through a main thread;
A method further comprising:
According to claim 1,
receiving, from the nodes of the cluster, a second gradient value related to the next iterated distributed learning;
applying the second gradient value to the learning model; and
joining the cluster;
A method further comprising:
In an apparatus for implementing a first node joining a cluster for distributed learning,
a transceiver for transmitting and receiving a signal;
Memory; and
an AI processor for functionally controlling the transceiver and the memory;
includes,
The AI processor
A learning model for distributed learning is received from the second node included in the cluster through the transceiver, and a background thread participates in the cluster for distributed learning, and the background thread is the distributed learning. Generates a tensor related to a gradient value for Applies values to the learning model.
7. The method of claim 6,
the second node
When the first node requests to join the cluster, the distributed learning is stopped in order to transfer the learning model to the first node.
7. The method of claim 6,
The tensor has the same type as the gradient value for distributed learning, and is initialized to 0.
7. The method of claim 6,
The AI processor
In order to apply the first gradient values obtained through the all-reduce operation to the learning model,
The apparatus of storing the obtained first gradient values in the memory, and sequentially applying the first gradient values stored in the memory to the learning model through a main thread.
7. The method of claim 6,
The AI processor
receiving, from a node of the cluster, a second gradient value associated with a next iterated distributed learning, applying the second gradient value to the learning model, and joining the cluster.

Images