Abstract: A method with neural network compression includes: generating a second neural network by fine-tuning a first neural network, which is pre-trained based on training data, for a predetermined purpose; determining delta weights corresponding to differences between weights of the first neural network and weights of the second neural network; compressing the delta weights; retraining the second neural network updated based on the compressed delta weights and the weights of the first neural network; and encoding and storing the delta weights updated by the retraining of the second neural network.

Abstract: A processor-implemented method with multi-task processing includes: obtaining weights of a first neural network; obtaining first delta weights of a second neural network that is fine-tuned from the first neural network, based on a target task; performing an operation of the second neural network on first input data, based on sums of the weights of the first neural network and the first delta weights; obtaining second delta weights of a third neural network that is fine-tuned from the first neural network, based on a change of the target task; replacing the first delta weights with the second delta weights; and performing an operation of the third neural network on second input data, based on sums of the weights of the first neural network and the second delta weights, wherein the first delta weights comprise difference values in the weights of the first neural network and weights of the second neural network, and the second delta weight comprises difference values in the weights of the first neural network an

Abstract: The present disclosure relates to a graphene spin transistor for all-electrical operation at room temperature and a logic gate using the graphene Rashba spin transistor. A graphene spin transistor of the present disclosure provides a graphene spin FET (Field Effect Transistor) for all-electrical operation at room temperature without a magnetic field or a ferromagnetic electrode by utilizing the Rashba-Edelstein effect in the graphene or the spin Hall effect of a TMDC (Transition Metal Dichalcogenide) material in order to replace CMOS transistors and extend Moore's Law, and further provides a logic gate using the graphene Rashba spin transistor.

Abstract: Various embodiments relate to an electronic device for supporting a finger pinch interaction using a return loss of a radio frequency signal and an operating method thereof. The electronic device is configured to measure a return loss of an antenna while radiating a radio frequency signal through the antenna and to detect whether a thumb and a finger touch each other within a predetermined distance from the antenna based on the measured return loss. Accordingly, there can be provided an interaction for the touch between the thumb and the finger.

Abstract: The present invention relates to a frequency modulation method based on a phase-locked loop capable of performing fast modulation independent of bandwidth. A frequency modulation system based on a phase-locked loop capable of performing fast modulation independent of bandwidth according to the present invention includes a loop filter including a proportional path and an integral path to determine a bandwidth of a phase-locked loop, a voltage-controlled oscillator configured to adjust a frequency according to an output of the loop filter, and a slope alternator configured to alternate an input current of the loop filter, wherein the slope alternator is located in the integral path of the loop filter to generate an offset current at a moment of change from a modulation rise to a modulation fall.

Abstract: The present invention relates to a transition metal nitride support, a method of manufacturing the same, a metal catalyst and a platinum-alloy catalyst including the transition metal nitride support, and manufacturing methods thereof. The manufactured transition metal support prevents corrosion of the support and aggregation of the platinum catalyst, thereby exhibiting high oxygen reduction catalytic activity. Also, strong metal-support interaction (SMSI) can be stabilized, thus improving the durability of the catalyst. The transition metal support includes large pores uniformly distributed therein, thereby increasing the amount of the catalyst supported and minimizing mass-transfer resistance in a membrane- electrode assembly, increasing the performance of a polymer electrolyte membrane fuel cell. The metal catalyst includes platinum particles loaded on the transition metal nitride support, thus exhibiting superior durability and activity.

Abstract: A method of feedback controlling a 3D printing process in real time, and a system therefor are disclosed. The method includes collecting big data, generated through 3D printing experiments, related to process variables of 3D printing, measurement signals, and 3D printing quality of the 3D printing object; building an artificial neural network model by performing machine-learning based on the collected big data; evaluating whether or not a 3D printing quality of the 3D printing object is abnormal in real time based on an actual measurement signal of the 3D printing object and the artificial neural network model; and feedback controlling printing quality of the 3D printing object in real time based on the evaluation result of whether or not the 3D printing quality of the 3D printing object is abnormal.

Abstract: A method of generating a cryptographic algorithm includes generating at least one key table on the basis of a seed value; generating, by using a round tweak bit string and an input bit string that is input to one of input branches of a Feistel structure, a first transformation function converting the input bit string into a first random bit string having a length that extends beyond a length of the input bit string, generating a second transformation function converting a second random bit string generated by referencing the one or more key tables into a third random bit string having a length that is the same as the length of the input bit string, and generating a block cryptographic algorithm of a Feistel structure which includes a round function to which the one or more key tables, the first transformation function, and the second transformation function are applied.

Abstract: A method of generating a cryptographic algorithm according to one embodiment of the present disclosure includes generating one or more key tables on the basis of a seed value; generating a first transformation function that converts an input bit string, which is input to one of input branches of a Feistel structure, into a first random bit string having a length that extends beyond a length of the input bit string; generating a second transformation function that converts a second random bit string generated by referencing the one or more key tables into a third random bit string having a length that is the same as the length of the input bit string; and generating a block cryptographic algorithm of a Feistel structure which includes a round function to which the one or more key tables, the first transformation function, and the second transformation function are applied.

Abstract: Disclosed are an electrode including a polymer matrix and a catalyst including metal nanoparticles and a conductive polymer shell and, a method of preparing the same. According to various exemplary embodiments of the present invention, various hybrid nano-composites may be formed by a combination of other conductive polymers than P3HT with metal nanoparticles. For example, the method may include selectively disposing metal nanoparticles to a surface modified conductive polymer including a block copolymer of two or more types of conductive polymers.

Abstract: A method of operating a solid oxide fuel cell system may comprise contacting a cathode gas comprising oxygen with a heating element to produce a heated cathode gas, passing the heated cathode gas through a cathode of a solid oxide fuel cell stack to increase the temperature of the solid oxide fuel cell stack to an operation temperature and reduce the oxygen to oxygen anions, and passing an anode gas through an anode of the solid oxide fuel cell stack to initiate the electrochemical oxidation of the oxygen anions within the anode. The passing of the anode gas through the anode of the solid oxide fuel cell stack may be initiated when the solid oxide fuel cell stack is heated to an operational temperature.

Abstract: The present invention relates to a dry reforming catalyst in which an active material is impregnated on the surface of a metal oxide support and the active material is surrounded by a surfactant, a method of preparing the same, and a method of producing a synthetic gas using the catalyst. Since the surfactant on the surface of the active material prevents the active material from being sintered and the active material surface from being covered with carbon, the dry reforming catalyst exhibits high activity at high temperature for a long period of time without having to use a precious metal, and thus is useful for the production of a synthetic gas.

Abstract: A modular fluid chip according to an embodiment of the present disclosure includes a body including at least one first hole which allows fluid to flow therethrough; and a housing receiving the body therein, and including a second hole which corresponds to the at least one first hole and allows the fluid to flow therethrough, and a fluid connection part which is connectable to another modular fluid chip.

Abstract: Disclosed are an anode for a lithium secondary battery, a lithium secondary battery including the anode, and a manufacturing method thereof. In particular, the anode includes a lithium metal layer and an interfacial layer made of phosphorous-doped graphitic carbon nitride and a single ion conducting polymer.

Abstract: A method and an apparatus for polarization multiplexing and encoding in a visible light communication system based on a gold nanoparticle are provided. A method for encoding in a receiver using a polarizing characteristic in a visible light communication system based on a gold nanoparticle, which is performed by a computer device, includes splitting a receive signal into left circular polarization light and right circular polarization light by allowing the receive signal to pass through a linear polarizer and estimating a phase retardation difference, and detecting a transmit symbol by estimating the phase retardation difference.

Abstract: An electronic device to co-locating models and a method of operating the electronic device is provided. The electronic device includes one or more of processors configured to analyze computational characteristics in response to a plurality of models being located to an accelerator, determine an affinity representing a utilization of the accelerator in response to two models among the plurality of models being co-located based on the computational characteristics of the plurality of models, and co-locate the two models among the plurality of models to the accelerator based on the affinity.

Abstract: According to an embodiment, in a solution provision method and system: a target solution for a target factory is determined in a platform in which solutions for a smart factory, the solutions including at least one template, are registered; at least one target template corresponding to the target solution is revised according to an environment of the target factory; and the target solution, to which the revised target template is applied, is applied to the target factory.

Abstract: Disclosed is an energy-efficient retraining method of a generative neural network for domain-specific optimization, including (a) retraining, by a mobile device, a pretrained generative neural network model with respect to some data of a new user dataset, (b) comparing, by the mobile device, the pretrained generative neural network model and a generative neural network model retrained for each layer with each other in terms of a relative change rate of weights, (c) selecting, by the mobile device, specific layers having high relative change rate of weights, among layers of the pretrained generative neural network model, as layers to be retrained, and (d) performing, by the mobile device, weight update for only the layers selected in step (c), wherein only some of all layers are selected and trained in a retraining process that requires a large amount of operation, whereby rapid retraining is performed in the mobile device.

Abstract: Disclosed is a method and apparatus for real-time estimation of full parameters of a permanent magnet synchronous motor. According to this method and apparatus, it is possible to estimate in real time all four parameters of a permanent magnet synchronous motor without additional signal injection. In addition to the state equation, the “stator current ripple model” is additionally used to fundamentally solve the rank deficiency problem in the state equation without injecting additional signals. All four parameters of a permanent magnet synchronous motor can be estimated in real time.

Abstract: Disclosed is an apparatus and method for operating vehicular accessories during an engine stop using a one-way clutch pulley, the apparatus including the one-way clutch pulley configured to drive the vehicular accessories, and including a coupler and a one-way clutch. The one-way clutch pulley includes an engine crank shaft configured to connect to an engine of a vehicle; the coupler configured to couple with the engine crank shaft; the one-way clutch configured to couple with the coupler; and a pulley configured to connect to the one-way clutch, and the one-way clutch is configured to connect to the pulley and to rotate in one direction.