Abstract: A method of recognizing an object based on mask updates is provided. The method includes generating first result information in which a foreground and background of a predetermined input image have been separated from each other, generating second result information in which the results of the object recognition within the input image and a boundary portion of the object have been matched, calculating an objective function to be applied to each mask head based on the first and second result information and parameter information for the first and second result information, updating all of the mask heads based on the objective function, and performing object recognition based on the updated mask.

Abstract: Disclosed are a method and apparatus for connecting a terminal and a base station in a multi-hop network. A method of a terminal may comprise: receiving cell selection and reselection criteria information broadcast from a base station; determining candidate base stations among all base stations capable of communicating with the terminal based on the cell selection and reselection criteria information; selecting one connection scheme from a lightening backhaul load (LBL) scheme or a maximum reference signal received power (RSRP) scheme based on a tolerable gap, which is a parameter for determining the candidate base station; selecting one candidate base station from among the candidate base stations as a target base station based on the one connection scheme; and connecting to the target base station.

Abstract: The present disclosure relates to a question answering system and method capable of inferring multiple correct answers. The question answering system capable of inferring multiple correct answers according to the present disclosure includes an input interface device configured to receive a query, a memory for storing a program that analyzes the query and searches for a paragraph with a high probability of including a correct answer through a document search, and a processor for executing the program, wherein the processor extracts a correct answer using a result of the searching for a paragraph with a high probability of including a correct answer, determines whether the extracted correct answer corresponds to multiple correct answers, and provides a correct answer extraction result.

Abstract: Proposed is an apparatus for performing high-capacity wireless communication in a greenhouse environment, the apparatus including: a moving enabling control unit including a motor for moving at least one wireless router in an upper end portion of a greenhouse, a moving enabling module including a motor control module and a motor communication module, and a moving enabling rails installed in a predetermined number, length and direction at the upper end portion of the greenhouse, and a greenhouse integrated control unit configured to set an interval of movement of the wireless router by communicating with the motor communication module and to generate and transmit a control command for controlling the motor control module in such a manner that the wireless router moves at the set interval of movement.

Abstract: Provided is a method of generating direction vectors of particles. The method includes calculating, by a processor, a center of gravity of the particles on the basis of characteristic values of one or more particles in a cluster including the particles, and generating, by the processor, direction vectors of the particles on the basis of the center of gravity.

Abstract: A method of a first communication node may comprise: transmitting, to a second communication node, an initiation signal indicating to initiate a collection procedure of training data for a neural network; transmitting, to the second communication node, an information signal including network parameters of the first communication node; receiving, from the second communication node, the training data in response to the information signal; training the neural network using the training data; determining optimal network parameters using the trained neural network; and performing communication with the second communication node using the optimal network parameters.

Abstract: Disclosed herein is a method for task planning for collaboration of artificial intelligence (AI) agents. The method includes generating a scene graph using an image acquired by an AI agent and a human instruction and generating a machine instruction set for objects in the scene graph, and the scene graph includes relevance information between each of the objects in the scene graph and the human instruction.

Abstract: A transparent structure according to an embodiment of the present disclosure includes a first panel and a second panel, wherein the first panel includes: a first electrode; a second electrode separated from the first electrode; and a polymer dispersed liquid crystal layer disposed between the first electrode and the second electrode, the second panel includes: a third electrode; a fourth electrode separated from the third electrode; and a colloid layer disposed between the third electrode and the fourth electrode, the fourth electrode includes multiple openings, and the colloid layer includes charged particles of which surfaces exhibit (+) charge or (?) charge.

Abstract: The present invention relates to verifying the suitability of a tool and diagnosing a spindle for a machining center on which different tools are mounted. Provided are a machining center spindle diagnosis apparatus and method configured to monitor a change of a tool in the machining center; when the change of the tool is recognized, control the machining center to idle the spindle; acquire sensor data from a sensor installed on the machining center during the idling of the spindle; input the acquired sensor data to a tool verifying model pre-trained by a machine learning technology to verify suitability of the tool; and when the tool is verified to be suitable, inputting the acquired sensor data to a spindle diagnosing model pre-trained by the machine learning technology to diagnose an operating state of the spindle.

Abstract: The present invention relates to a system and method for predicting AI useful life based on accelerated life testing data. The system for predicting AI useful life based on accelerated life testing data according to the present invention includes a feature extraction unit configured to receive accelerated life training data and actual operation testing result and encodes the received accelerated life training data and actual operation testing result into a latent variable, a regression network configured to be branched for each domain of data received by the feature extraction unit, and a domain discrimination network configured to map the accelerated life training data and actual operation testing result to the latent variables in a latent space.

Abstract: A network device divided into a training plane and a control plane, model management server that controls a network device, and processing methods of a network device and model management server are disclosed. A processing method may include receiving a machine learning model from a model management server, obtaining network data to generate analytics information, generating analytics information by inputting the network data to a machine learning model, feeding back the analytics information to the model management server, and generating a control command of the network device using the analytics information, wherein the analytics information is generated by a training plane function and the control command is generated by a control plane function.

Abstract: Disclosed herein is a method for distributed training of an AI model in a channel-sharing network environment. The method includes determining whether data parallel processing is applied, calculating a computation time and a communication time when input data is evenly distributed across multiple computation devices, and unevenly distributing the input data across the multiple computation devices based on the computation time and the communication time.

Abstract: Disclosed herein are a method for machine-learning parallelization using host CPUs of a multi-socket structure and an apparatus therefor. The method, performed by the apparatus for machine-learning parallelization using host CPUs of a multi-socket structure, includes a compile phase in which a learning model is split at a layer level for respective pipeline stages and allocated to Non-Uniform Memory Access (NUMA) nodes for respective CPU sockets and a runtime phase in which parameters required for learning are initialized and multiple threads generated in consideration of a policy of each parallelism algorithm are executed by being allocated to respective cores included in the NUMA node.

Abstract: A method for verifying reliability of an artificial intelligence (AI) model includes receiving an AI model request; creating a verification twin for evaluating the reliability of the AI model; and verifying the reliability of the AI model based on information collected while the AI model is executed on the digital twin network.

Abstract: Disclosed herein are an artificial intelligence application provision method and apparatus for supporting Edge computing for Cyber-Physical Systems (EdgeCPS). The artificial intelligence application provision method includes receiving an artificial intelligence application and service specification, obtaining artificial intelligence-related information allocated from an artificial intelligence information sharing database based on the artificial intelligence application and service specification, creating a pipeline specification corresponding to the artificial intelligence application and service specification, and allocating resources corresponding to respective pipelines using the pipeline specification.

Abstract: Disclosed herein are a quantum simulation apparatus and method using logical qubit synthesis. The quantum simulation apparatus is configured to generate a new virtual quantum device configured using magic qubits by dividing each logical qubit into multiple areas, extract logical qubit information from a target virtual quantum device that is a target of a logical Controlled NOT (CNOT) operation, synthesize the extracted logical qubit information into the new virtual quantum device, perform a logical CNOT operation in the synthesized virtual quantum device, extract logical qubit information from a result of the logical CNOT operation in the synthesized virtual quantum device, and synthesize the logical qubit information extracted from the result of the logical CNOT operation into the target virtual quantum device.

Abstract: An apparatus for generating text from an image may comprise: a memory configured to store at least one instruction; and a processor configured to execute the at least one instruction, wherein the processor is further configured to generate encoding information for an image based on the image and extract text information related to content of the image based on a degree of association with the encoding information.

Abstract: An identification method and an unmanned aerial vehicle (UAV) detection apparatus are provided. The identification method, as a method for preventing a collision that may occur in a process of detecting and identifying a plurality of UAVs flying in the air, detects a plurality of UAVs in the air and determines a legitimacy of flight of each of the plurality of UAVs using an identification information request message.

Abstract: An encoding method obtains an intra prediction mode related to a current block from a bitstream; obtains prediction samples of the current block by performing intra prediction on the current block based on the intra prediction mode; scans transform coefficients of the current block, obtains dequantized transform coefficients of the current block by performing a dequantization on the transform coefficients; determines a transform type of the current block; and obtains residual samples of the current block by performing an inverse-transform on the dequantized transform coefficients based on the determined transform type, the inverse-transform being performed in a horizontal direction and a vertical direction. The transform type is determined to be a Discrete Cosine Transform or Discrete Sine Transform, the size of the current block is equal to or greater than 4×4, and the transform type is determined independently to a direction of the intra prediction mode of the current block.

Abstract: Disclosed herein a disaggregation computing system. The disaggregation computing system comprising: a local computing device that comprises a local processor, a local memory bus, a local memory and a local disaggregation controller; a remote computing device that comprises a remote processor, a remote memory bus, a remote memory and a remote disaggregation controller; and a disaggregation network that connects the local computing device and the remote computing device, wherein the local disaggregation controller and the remote disaggregation controller are configured to: check a response delay for access of the remote memory, and control the access of the remote memory based on the response delay.