Abstract: An inference system applies a machine-learning transformer model to a batch of requests with variable input length or variable target length or variable internal state length by selectively batching a subset of operations in the transformer model but processing requests in the batch individually for a subset of operations in the transformer model. In one embodiment, the operation to be processed individually is an attention operation of an encoder or a decoder of the transformer model. By selective batching, the inference system can allow batching operations to be performed for a batch of requests with variable input or target length or internal state length to utilize the parallel computation capabilities of hardware accelerators while preventing unnecessary computations that occur for workarounds that restrain the data of a batch of requests to a same length.

Abstract: Provided are a method of treating chronic kidney disease or one or more symptoms thereof, or reducing a complication related to chronic kidney disease, the method comprising, administering a therapeutically effective amount of gamma-cyclodextrin oligomer to a subject in need thereof. The gamma-cyclodextrin oligomers are effective in cholesterol metabolism enhancement, cholesterol efflux, reducing inflammatory cytokine secretion, renal clearance of cholesterol and/or reducing albuminuria. Therefore, the gamma-cyclodextrin oligomers can be used to treat or alleviate chronic kidney disease, symptoms thereof and/or a complication related to chronic kidney disease.

Abstract: A positive electrode and a secondary battery including the same are provided. The positive electrode includes a current collector and a positive electrode active material layer disposed on the current collector, wherein the positive active material layer includes a positive electrode active material, a binder, and a multi-walled carbon nanotube, wherein the multi-walled carbon nanotube has an average length of 1-2 ?m and has a length standard deviation of 0.5 ?m or less.

Abstract: An image encoding/decoding method is provided. An image decoding method of the present invention may comprise deriving an intra-prediction mode of a current luma block, deriving an intra-prediction mode of a current chroma block based on the intra-prediction mode of the current luma block, generating a prediction block of the current chroma block based on the intra-prediction mode of the current chroma block, and the deriving of an intra-prediction mode of a current chroma block may comprise determining whether or not CCLM (Cross-Component Linear Mode) can be performed for the current chroma block.

Abstract: A quantum dot including a core and a shell disposed on the core wherein one of the core and the shell includes a first semiconductor nanocrystal including zinc and sulfur and the other of the core and the shell includes a second semiconductor nanocrystal having a different composition from the first semiconductor nanocrystal, the first semiconductor nanocrystal further includes a metal and a halogen configured to act as a Lewis acid in a halide form, an amount of the metal is greater than or equal to about 10 mole percent (mol %) based on a total number of moles of sulfur, and an amount of the halogen is greater than or equal to about 10 mol % based on a total number of moles of sulfur, a method of producing the same, and a composite and an electronic device including the same.

Abstract: An electrode includes an electrode active material, wherein the electrode active material layer includes an electrode active material, polyvinylidene fluoride, and a conductive agent, wherein the conductive agent includes a carbon nanotube structure in which 2 to 5,000 single-walled carbon nanotube units are bonded to each other, and the carbon nanotube structure is included in an amount of 0.01 wt % to 0.5 wt % in the electrode active material layer. A secondary battery including the same, and a method of preparing the electrode are also provided.

Abstract: An adhesive film, an optical member including the same, and an optical display apparatus including the same are provided. An adhesive film includes a (meth)acrylic based copolymer, a (meth)acrylic based oligomer, and a curing agent and has a maximum shear strain of 50% or less at 60° C. and a peel strength of 600 gf/25 mm or more with respect to a polyimide based film at 25° C.

Abstract: The present invention relates to a method for preparing an alkali metal ion conductive chalcogenide-based solid electrolyte, a solid electrolyte prepared thereby, and an all-solid-state battery comprising the same.

Abstract: An emissive display device includes a display area which includes a plurality of pixels, and a driver disposed at a side of the display area, where the driver includes at least two emission signal stages disposed in one row, and an input signal line connected to the emission signal stages, and the at least two emission signal stages are connected to the same input signal line.

Abstract: An apparatus and method for manufacturing iron-based mixed powder with excellent flowability is provided. The apparatus includes a hopper which stores and discharges a main raw material of iron-based powder, a transport means which transports the main raw material of iron-based powder discharged from the hopper, a magnetizing means that applies magnetic force to the main raw material transported and falling from the transport means to process the main raw material of iron-based powder into a main raw material bundle in a crumbly type in which the main raw material of iron-based powder is agglomerated with each other, a first mixer in which the main raw material bundle in a magnetized state and an auxiliary raw material of iron-based powder are loaded and mixed while being rotated and transported, and a second mixer in which a first iron-based mixed powder is mixed while being rotated and transported.

Abstract: An inference system applies a machine-learning transformer model to a batch of requests with variable input length or variable target length or variable internal sate length by selectively batching a subset of operations in the transformer model but processing requests in the batch individually for a subset of operations in the transformer model. In one embodiment, the operation to be processed individually is an attention operation of an encoder or a decoder of the transformer model. By selective batching, the inference system can allow batching operations to be performed for a batch of requests with variable input or target length or internal state length to utilize the parallel computation capabilities of hardware accelerators while preventing unnecessary computations that occur for workarounds that restrain the data of a batch of requests to a same length.

Abstract: A printed circuit board includes a first insulating layer; a through-hole penetrating a region between an upper surface and a lower surface of the first insulating layer; a first via disposed in at least a portion of the through-hole; a first pad connected to an upper side of the first via; and a second pad connected to a lower side of the first via. A width of the through-hole on an uppermost side is greater than a width of the first pad. At least a portion of an upper surface of the first via exposed from the first pad is recessed below the upper surface of the first insulating layer.

Abstract: The present invention relates to a method and device for diagnosing a defect of a collaborative robot, the method comprising the steps in which: an electronic device generates a sensing data structure for managing sensing data collected from at least one collaborative robot; the electronic device generates an operation data structure for managing operation data associated with the operation of the collaborate robot; the electronic device generates a malfunction data structure for managing malfunction data of a point in which the severity of the operation equals to or is higher than a threshold; and the electronic device stores data collected from the at least one collaborative robot in accordance with the structures. Application to other embodiments is also possible.

Abstract: An electronic device includes a first substrate, a plurality of light emitting elements each having a horizontal length and a vertical length which are less than or equal to about 10 micrometers (?m), each of the plurality of light emitting elements being disposed on the first substrate, a quantum dot color filter layer disposed on the plurality of light emitting elements, and a first overcoat layer between a plurality of light emitting elements and the quantum dot color filter layer. The quantum dot color filter layer includes a plurality of quantum dot color filters partitioned by a plurality of first partition walls so as to be overlapped with the plurality of light emitting elements, respectively.

Abstract: Disclosed herein are a video decoding method and apparatus and a video encoding method and apparatus. In video encoding and decoding, multiple partition blocks are generated by splitting a target block. A prediction mode is derived for at least a part of the multiple partition blocks, among the multiple partition blocks, and prediction is performed on the multiple partition blocks based on the derived prediction mode. When prediction is performed on the partition blocks, information related to the target block may be used, and information related to an additional partition block, which is predicted prior to the partition block, may be used.

Abstract: A quantum dot-polymer composite pattern including at least one repeating section configured to emit light of a predetermined wavelength, and a production method and a display device including the quantum dot-polymer composite are disclosed. The quantum dot-polymer composite includes a polymer matrix including linear polymer including a carboxylic acid group-containing repeating unit and a plurality of cadmium-free quantum dots dispersed in the polymer matrix, has an absorption rate of greater than or equal to about 85% for light at wavelength of about 450 nm, and has an area ratio of a hydroxy group peak relative to an acrylate peak of greater than or equal to about 2.6 in Fourier transform infrared spectroscopy.

Abstract: There is provided an image encoding/decoding method and apparatus. An image decoding method according to the present invention may comprise reconstructing a most frequent mode (MFM) indicator indicating whether or not an intra prediction mode of a current block is a predetermined intra prediction mode, deriving, on the basis of the MFM indicator, the intra prediction mode of the current block, and generating a prediction block of the current block by performing intra prediction on the current block on the basis of the intra prediction mode.

Abstract: The invention encompasses methods and test strips for detecting the presence of cerebrospinal fluid (CSF) in a biological sample with a lateral flow device which uses lectin conjugates, anti-antigen conjugates, an immobilized serum line, and an immobilized anti-antigen line.

Abstract: A dual slot die coater according to the present disclosure is a dual slot die coater including a lower slot and an upper slot, for extrusion coating of an electrode active material slurry on a surface of a continuously moving current collector through at least one of the lower slot or the upper slot, and the dual slot die coater includes a lower plate, an intermediate plate positioned on the lower plate and an upper plate positioned on the intermediate plate, the lower slot being formed between the lower plate and the intermediate plate, and the upper slot being formed between the intermediate plate and the upper plate, wherein the lower plate, the intermediate plate and the upper plate have a lower die lip, an intermediate die lip and an upper die lip, each forming an front end with respect to the current collector, respectively, and a thickness of the lower die lip is larger than a thickness of the upper die lip and a thickness of the intermediate die lip.

Abstract: The present invention provides a transformer malfunction diagnostic device and a malfunction diagnosis method using same, wherein a rule-based learning method is combined with a deep-learning-based learning method based on artificial intelligence. The transformer malfunction diagnosis device according to an embodiment of the present invention comprises: a data scaling unit for scaling dissolved gas analysis data acquired from a transformer, a provisionally labeled data acquisition unit for converting unlabeled data, among the scaled dissolved gas analysis data, to provisionally labeled data and acquiring same; a prelearning unit for performing prelearning for the unlabeled data, among the scaled dissolved gas analysis data, and the provisionally labeled data; and a relearning unit for performing relearning for a labeled data through the transformer of parameters optimized by performing the prelearning.