Abstract: An apparatuses for managing access to a file stored in a remote location includes detecting an attempt for a user terminal to access the file stored in the remote location, determining a location where the file attempted to be access is to be opened, in response to a determination that the location where the file is to be opened is remote, opening the file and providing a content of the opened file to the user terminal by a file management server and in response to a determination that the location where the file is to be opened is local, processing the file to be downloaded to the user terminal by the file management server.

Abstract: A method for managing an application in a short-range communication device according to some embodiments of the present disclosure includes receiving application installation request information from an administrator of the short-range communication device, generating authentication information for installing the application by using the received application installation request information, requesting a first server to verify the authentication information, downloading application installation data based on a verification of the authentication information, and installing the application in an application platform region of the short-range communication device using the application installation data, and installing an applet for interworking with the application in a secure element of the short-range communication device.

Abstract: A blockchain based data management method performed by a computing device according to an embodiment of the present disclosure includes recording a deletion event for off-chain data and time information of the deletion event in a blockchain network, and selectively deleting the off-chain data based on a validity verification result calculated for the time information, wherein the validity verification result is calculated using a time-consensus algorithm of the blockchain network.

Abstract: An apparatus is for unsupervised domain adaptation for allowing a deep learning model with supervised learning on a source domain completed to be subjected to unsupervised domain adaptation to a target domain. The apparatus includes a first learning unit to perform a forward pass by inputting a pair (xsi, ysi) of first data xsi of the source domain and a label ysi for each of the first data and second data xTj belonging to the target domain, and insert a dropout following a Bernoulli distribution into the deep learning model in performing the forward pass, and a second learning unit to perform a back propagation to minimize uncertainty about the learning parameter of the deep learning model by using a predicted value for each class output through the forward pass and the label ysi, and an uncertainty vector for the second data xTj output through the forward pass as inputs.

Abstract: A system for a remote support according an embodiment includes a web viewer configured to receive a request for execution of a remote support service from a logged-in administrator, a web application server configured to receive a request for generation of a connection space for the remote support service from the web viewer and generate a first connection space and a first connection code for identifying the first connection space based on the request for generation, a relay server configured to receive the first connection code, generate a second connection space in which the first connection code is stored, and receive a request for provision of the remote support service from a mobile terminal, and a message broker configured to mediate a plurality of message queues between the first connection space and the second connection space for the remote support service between the web viewer and the mobile terminal.

Abstract: A synthetic data generation apparatus according to an embodiment includes a generator for generating synthetic data from an input value, a first discriminator learned to distinguish between actual data and the synthetic data, a second discriminator learned to distinguish between the actual data and the synthetic data while satisfying differential privacy, and a third discriminator learned to distinguish between first synthetic data which is output from the generator learned by the first discriminator and second synthetic data which is output from the generator learned by the second discriminator.

Abstract: An apparatus for document recognition according to an embodiment includes a document type analyzer that analyzes a type of a recognition target document based on document feature vector extracted from one or more partial images obtained by color space conversion of one or more partial regions of the recognition target document, and an information extractor that extracts value information from one or more information search images organized in a grid form based on a position of key information of the recognition target document.

Abstract: A method performed by a computing device for controlling a P2P connection according to an embodiment of the present disclosure includes obtaining network information of a current path for transmitting and receiving data, calculating requested data quality of the current path by using the network information, and changing a network path to an alternative path by using previously stored data quality history information of the alternative path when a difference between a predetermined highest data quality and the requested data quality of the current path is equal to or greater than a reference value.

Abstract: A method for clustering based on unsupervised learning according to an embodiment of the invention enables clustering for newly generated patterns and is robust against noise, and does not require tagging for training data. According to one or more embodiments of the invention, noise is accurately removed using three-dimensional stacked spatial auto-correlation, and multivariate spatial probability distribution values and polar coordinate system spatial probability distribution values are used as learning features for clustering model generation, making them robust to noise, rotation, and fine unusual shapes. In addition, clusters resulting from clustering are classified into multi-level clusters, and stochastic automatic evaluation of normal/defect clusters is possible only with measurement data without a label.

Abstract: An apparatus for managing a security policy of a firewall according to an embodiment includes a rule request module that receives one or more requested rules to be applied to a firewall, a rule merge module that merges a pre-applied rule of the firewall and the one or more requested rules when the number of rules applied to the firewall exceeds a maximum number of rule registrations of the firewall due to the requested rule, and a firewall interface module that receives the pre-applied rule from the firewall and provides the pre-applied rule to the rule merge module, and re-registers a merged rule merged through the rule merge module in the firewall, and the rule merge module is configured to merge the pre-applied rule and the one or more requested rules so that a security vulnerable space occurring due to the merging is minimized.

Abstract: A method for training a deep neural network according to an embodiment includes training a deep neural network model using a first data set including a plurality of labeled data and a second data set including a plurality of unlabeled data, assigning a ground-truth label value to some of the plurality of unlabeled data, updating the first data set and the second data set such that the data to which the ground-truth label value is assigned is included in the first data set, and further training the deep neural network model using the updated first data set and the updated second data set.

Abstract: A blockchain-based data distribution management method according to an embodiment is performed by a computing device. The method includes generating a plurality of shards by fragmenting a first information, requesting a first block event for distributed storage of the plurality of shards in a plurality of off-chain storages to a blockchain network, and transmitting the plurality of shards to the plurality of off-chain storages, wherein a second information indicating a history of the plurality of shards distributed and stored in the plurality of off-chain storages is recorded in the blockchain network through the first block event, wherein when a predetermined number or more of shards among the plurality of shards are collected, the first information is restored from the collected predetermined number or more of shards.

Abstract: A data transmission method according to an embodiment is performed by a computing device including one or more processors and a memory in which one or more programs to be executed by the one or more processors are stored. The data transmission method includes receiving, from connected client devices, peer-to-peer (P2P) network information of the client devices, extracting one or more of information on whether network access is successful and network quality information, which are matched with the P2P network information and pre-stored, and determining one of a first data transmission path and a second data transmission path as a data transmission path for the client devices based on one or more of the information on whether network access is successful and the network quality information.

Abstract: An apparatus and method for analyzing time series data on the basis of machine learning are provided. According to the disclosed embodiments, it is possible to effectively augment time series data, which is a target to be learned, according to characteristics of the time series data, thereby solving a problem of overfitting a machine learning model due to limited training data and a problem of deterioration of prediction accuracy due to imbalance of distribution of time series data and improving reliability of time series data analysis. In addition, according to the disclosed embodiments, it is possible to effectively set an optimal parameter for augmenting time series data.

Abstract: A method for encryption according to an embodiment includes generating a ciphertext for a secret key that is an integer vector by using an integer-based first homomorphic encryption algorithm, generating a key stream that is the integer vector from a nonce and the secret key by using a key stream generator, encoding the key stream by using a message encoding function of the first homomorphic encryption algorithm, encoding a message that is a real vector by using a message encoding function of a real number-based second homomorphic encryption algorithm, generating a ciphertext for the message by using a result of the encoding of the key stream and a result of the encoding of the message, and transmitting the nonce, the ciphertext for the secret key, and the ciphertext for the message to an apparatus for converting a ciphertext.

Abstract: A computing device according to an embodiment includes one or more processors, a memory in which one or more programs to be executed by the one or more processors are stored, a queue manager configured to receive a security detection request including a detection target file, a plurality of detectors configured to perform a security detection operation on the target detection file, and an operation distributor configured to acquire detection policy information related to the detection target file and distribute the detection target file received from the queue manager to one or more of the plurality of detectors based on the detection policy information.

Abstract: A method for model customization according to an embodiment includes providing a user with prediction results of each of a plurality of pre-trained natural language processing models for a document subjected to analysis selected from a document set including a plurality of documents, acquiring user feedback on the prediction results from the user, generating a plurality of augmented documents from at least one of the plurality of documents based on data attributes of each of the plurality of documents and the user feedback; and retraining at least one of the plurality of natural language processing models, using training data including the plurality of augmented documents.

Abstract: A method for batch processing for a plurality of individual transactions includes generating a batch transaction by aggregating at least some of the plurality of individual transactions according to a setting value of a batch size, and processing the batch transaction via a blockchain network. The generation of the batch transaction includes adjusting the setting value of the batch size based on a monitoring result for a transaction processing status, and generating the batch transaction according to the adjusted setting value.

Abstract: A blockchain data recording method includes determining a recording type as one of an on-chain type and an off-chain type according to a target data to be recorded in a blockchain, transmitting a first transaction proposal for recording the target data in a blockchain node configured to store an on-chain data on a blockchain to the blockchain node when the recording type is an on-chain type, and recording the target data in a storage node provided to store an off-chain data, and transmitting a second transaction proposal for storing recording information of the target data in the blockchain to the blockchain node when the recording type is an off-chain type.

Abstract: A machine learning apparatus according to an embodiment includes a feature extractor configured to extract features from an object region of an image, a label processor configured to create sentence label embeddings from a sentence label corresponding to the object region, a first training data creator to extract first sub-features from a plurality of first sub-regions created by partitioning the object region, add the sentence label embeddings to the extracted first sub-features, and add the first sub-features added with the sentence label embeddings to the features of the object region, a second training data creator to extract a plurality of second sub-regions along a bounding surface of the object region, create an attention matrix from the second sub-regions, and create a training data by applying the attention matrix to the features of the object region, and a trainer to train an object detection model using the training data.