Abstract: A manufacturing system is disclosed herein. The manufacturing system includes one or more stations, a monitoring platform, and a control module. Each station of the one or more stations is configured to perform at least one step in a multi-step manufacturing process for a component. The monitoring platform is configured to monitor progression of the component throughout the multi-step manufacturing process. The control module is configured to dynamically adjust processing parameters of each step of the multi-step manufacturing process to achieve a desired final quality metric for the component.

Abstract: A post-processing apparatus includes storage that stores a multi-task learning model. The post-processing apparatus also includes a controller that may generate a first entropy image from an output image of a first deep neural network (DNN) within the multi-task learning model. The controller may also generate a second entropy image from an output image of a second DNN in the multi-task learning model. The controller may post-process the output image of the first DNN based on the first entropy image and the second entropy image.

Abstract: A method and a system for providing a roughness haptic sense of a virtual object by using space-time encoding are provided. The method for providing a roughness haptic sense of a virtual object may comprise the steps of: displaying virtual objects to a user; tracking the position of the user's hand; predicting a virtual object, which is most likely to collide with the hand, among the virtual objects; and when the hand touches a haptic sense providing device, controlling the haptic sense providing device on the basis of the surface roughness of the predicted virtual object to feed back a roughness haptic sense to the user.

Abstract: A manufacturing system is disclosed herein. The manufacturing system includes one or more stations, a monitoring platform, and a control module. Each station of the one or more stations is configured to perform at least one step in a multi-step manufacturing process for a component. The monitoring platform is configured to monitor progression of the component throughout the multi-step manufacturing process. The control module is configured to dynamically adjust processing parameters of each step of the multi-step manufacturing process to achieve a desired final quality metric for the component.

Abstract: An apparatus and a method are for generating a depth map of a monocular camera image. The apparatus includes: a camera sensor that photographs a monocular camera image; a lidar sensor that generates a point cloud corresponding to the monocular camera image; and a controller that removes noise from the point cloud and generates the depth map of the monocular camera image based on the point cloud from which the noise is removed.

Abstract: Disclosed is an apparatus and method for controlling a turn-on sequence of switches included in a parallel multi battery pack, including a processor that receives a switch voltage value from a voltage sensor coupled to a kth sequence switch unit (k is a turn-on sequence index, which is a natural number of 2 to n) before turning on the kth sequence switch, then turns on the kth sequence switch, subsequently receives a switch current value from a current sensor connected to the kth sequence switch, counts a watt energy value consumed in the kth sequence switch from the switch voltage value and the switch current value and determines the turn-on sequence of the first to nth switches in descending order from a switch having a greatest watt energy count value.

Abstract: An active learning device and an active learning method are provided. The active learning device includes an input device configured to receive an input image, and a controller configured to extract a first feature from the input image, extract a second feature from a segmentation image corresponding to the input image, and extract a third feature from a depth image corresponding to the input image. The active learning device is also configured to determine the input image as a training image based on cosine distances among the first feature, the second feature and the third feature, and labels the training image.

Abstract: Provided are an apparatus for collecting a training image of a deep learning model and a method for the same. The apparatus may include a camera sensor configured to capture an image of a surrounding of a vehicle and a controller configured to determine an entropy of each pixel in the image. The controller may also be configured to determine an entropy of each class in the image based on the entropy of each pixel in the image and determine whether to collect the image based on the entropy of each class.

Abstract: An electronic device includes a communication circuitry, a microphone, a speaker, a memory, and a processor. The processor may identify a first audio signal received through the microphone. The processor may perform, based on identifying that the first audio signal may include a speech for sequentially executing a plurality of functions, at least one first function from among the plurality of functions. The processor may perform, based on identifying that the first audio signal may include the speech, at least one second function from among the plurality of functions on an external electronic device through the communication circuitry, the at least one second function being different from the at least one first function. The processor may sequentially output, through the speaker, at least one second audio signal representing at least one result of performing the at least one first function and the at least one second function based on a sequence associated with the speech.

Abstract: An apparatus for selecting a training image of a deep learning model and a method thereof are disclosed. The apparatus includes an input device and a controller. The input device receives a simulation image and information about an object in the simulation image from a simulation tool and receives a training image corresponding to the simulation image from an image conversion device. The controller detects a similarity between a structure of the object in the simulation image and a structure of an object in the training image and determines validity of the training image based on the detected similarity.

Abstract: An apparatus for processing a deep learning model includes a first memory, a second memory, and a controller. The controller is configured to, for each layer of the deep learning model, detect memory usage, a first processing time corresponding to the first memory being used, and a second processing time corresponding to the second memory being used, and determine an optimal memory for each layer of the deep learning model based on an objective function.

Abstract: An apparatus and method for detecting performance degradation of an autonomous vehicle include a storage configured to store a deep learning model used for image recognition in the autonomous vehicle and a controller configured to determine entropy of each pixel among a plurality of pixels in an output image of the deep learning model, determine an average value of the entropy of the plurality of pixels, and detect performance degradation of the deep learning model based on the average value of the entropy of the plurality of pixels.

Abstract: Disclosed herein are a reasoning method based on a structural attention mechanism for knowledge-based question answering and a computing apparatus for performing the reasoning method. The reasoning method includes: recognizing one or more entities in a query including content and a question, and linking the recognized entities to a knowledge base; constructing a question hypergraph and a query-aware knowledge hypergraph by performing a multi-hop graph walk on a question graph and the knowledge base; and inferring a correct answer to the question by applying as attention mechanism to a query hyperedge and a knowledge hyperedge included in the question hypergraph and the query-aware knowledge hypergraph, respectively.

Abstract: The present invention relates to a carbon dioxide conversion process and, more particularly, to a continuous carbon dioxide conversion process and a system therefor.

Abstract: Provided are a fusion protein including glutathione-S-transferase and a protein having binding affinity for a target cell or a target protein, and use thereof as a drug delivery carrier and a pharmaceutical composition. The fusion protein and the drug delivery carrier including the same according to an aspect may sustain an in vivo residence time, and may also have improved target cell-targeting ability, and thus may be effectively delivered to target cells. Accordingly, it may be usefully applied to a targeted therapeutic agent.

Abstract: An apparatus and method for diagnosing a failure of a switch unit included in a multi battery pack. The magnitude of a pack current flowing from each of battery packs connected in parallel to a load is measured, a battery pack whose pack current has a magnitude smaller than the average value by a threshold value or more is identified, and it is determined that an open failure occurs at the switch unit included in the identified battery pack, and the information of the battery pack including the switch unit at which the open failure occurs is output. The internal resistance of each battery pack may be statistically analyzed in order to more accurately diagnose the failure at the switch unit. Thus, an open failure of switch units included in battery packs connected in parallel may be easily and reliably diagnosed in a statistical viewpoint.

Abstract: Aspects of the disclosed technology provide a computational model that utilizes machine learning for detecting errors during a manual assembly process and determining a sequence of steps to complete the manual assembly process in order to mitigate the detected errors. In some implementations, the disclosed technology evaluates a target object at a step of an assembly process where an error is detected to a nominal object to obtain a comparison. Based on this comparison, a sequence of steps for completion of the assembly process of the target object is obtained. The assembly instructions for creating the target object are adjusted based on this sequence of steps.

Abstract: Methods for forming hafnium oxide within a three-dimensional structure, such as in a high aspect ratio hole, are provided. The methods may include depositing a first hafnium-containing material, such as hafnium nitride or hafnium carbide, in a three-dimensional structure and subsequently converting the first hafnium-containing material to a second hafnium-containing material comprising hafnium oxide by exposing the first hafnium-containing material to an oxygen reactant. The volume of the second hafnium-containing material may be greater than that of the first hafnium-containing material. Voids or seams formed during the deposition of the first hafnium-containing material in the three-dimensional structure may be filled by the expanded material after exposing the first hafnium-containing material to the oxygen reactant. Thus, the three-dimensional structure, such as a high aspect ratio hole, can be filled with hafnium oxide substantially free of voids or seams.

Abstract: Aspects of the disclosed technology provide an Artificial Intelligence Process Control (AIPC) for automatically detecting errors in a manufacturing workflow of an assembly line process, and performing error mitigation through the update of instructions or guidance given to assembly operators at various stations. In some implementations, the disclosed technology utilizes one or more machine-learning models to perform error detection and/or propagate instructions/assembly modifications necessary to rectify detected errors or to improve the product of manufacture.