Abstract: The present disclosure relates to an antibody to inhibit function of Angiopoietin-2 (Ang-2) by binding specifically to Ang-2, and directed to an anti-Ang2 antibody, a nucleic acid encoding the same, a vector comprising the nucleic acid, a cell transformed with the vector, a method of preparing the same, an angiogenesis inhibitor comprising the same, a composition for treating a disease related with Angiopoietin-2 activation and/or overproduction, a composition for diagnosing a disease related with Angiopoietin-2 activation and/or overproduction, a composition for treating eye disease or a composition for preventing or treating a cancer, and a composition for combining an antibody binding to Ang2 with a drug other than the antibody binding to Ang2.

Abstract: A method of generating a fixed-point type neural network by quantizing a floating-point type neural network, includes obtaining, by a device, a plurality of post-activation values by applying an activation function to a plurality of activation values that are received from a layer included in the floating-point type neural network, and deriving, by the device, a plurality of statistical characteristics for at least some of the plurality of post-activation values. The method further includes determining, by the device, a step size for the quantizing of the floating-point type neural network, based on the plurality of statistical characteristics, and determining, by the device, a final fraction length for the fixed-point type neural network, based on the step size.

Abstract: One embodiment of a camera module comprises: a lens barrel including at least one lens; a bobbin for accommodating the lens barrel; and a fixing part disposed between the lens barrel and the bobbin so as to inhibit the lens barrel from being separated from the bobbin, wherein one surface, which makes surface contact with the fixing part, of the bobbin can include a bobbin provided so as to protrude toward the lens barrel.

Abstract: A method of predicting a shape of a semiconductor device includes implementing a modeled semiconductor shape with respect to a designed semiconductor layout, extracting a plurality of samples by independently linearly combining process variables with respect to the modeled semiconductor shape; generating virtual spectrums with respect to ones of the extracted plurality of samples through optical analysis, indexing the virtual spectrums to produce indexed virtual spectrums, generating a shape prediction model by using the indexed virtual spectrums as an input and the modeled semiconductor shape as an output, and indexing a spectrum measured from a manufactured semiconductor device and inputting the spectrum to the shape prediction model to predict a shape of the manufactured semiconductor device.

Abstract: A method of quantizing an artificial neural network includes dividing an input distribution of the artificial neural network into a plurality of segments, generating an approximated density function by approximating each of the plurality of segments, calculating at least one quantization error corresponding to at least one step size for quantizing the artificial neural network, based on the approximated density function, and determining a final step size for quantizing the artificial neural network based on the at least one quantization error.

Abstract: A SCR catalyst for removing nitrogen oxides comprises: a carrier prepared from a support in which Ti-PILC is mixed with titania; and a catalyst material on the carrier, wherein the catalyst material contains an active material of a vanadium component and a co-catalyst of a tungsten component. On the basis of the total weight of the catalyst, the support Ti-PILC is contained at 0.01-40 wt %, and the support titania is contained at 50 to 90 wt %. In addition, a method for producing a SCR catalyst for removing nitrogen oxides comprises the steps of: preparing a carrier by using a support in which Ti-PILC is mixed with titania; and supporting a catalyst material on the carrier. The present disclosure provides: a SCR catalyst for removing nitrogen oxides, which has an excellent nitrogen oxide removing performance and a high producing convenience; and a method for producing the same.

Abstract: One embodiment comprises a circuit board including a mounting groove, an image sensor arranged within the mounting groove of the circuit board, and a first epoxy arranged within the mounting groove, wherein the mounting groove comprises a first side surface and a second side surface that face each other, and a third side surface and a fourth side surface that face each other, the circuit board includes at least one application groove provided on the first side surface and/or the second side surface of the mounting groove, the at least one application groove includes an opening opened toward the upper surface of the circuit board, and at least a portion of the first epoxy is arranged in the at least one application groove.

Abstract: Disclosed are an anti-PD-L1 antibody or an antigen-binding fragment thereof, a nucleic acid encoding the same, a vector including the nucleic acid, a cell transformed with the vector, a method for producing the antibody or an antigen-binding fragment thereof, a pharmaceutical composition for preventing or treating cancer containing the same, and a composition for co-administration with an antibody other than an antibody binding to PD-L1, containing the same.

Abstract: The present invention relates to an anti-VISTA antibody or an antigen binding fragment thereof, a nucleic acid coding for the same, a vector carrying the nucleic acid, a cell transformed with the vector, a method for preparing the antibody or an antigen binding fragment thereof, a composition for prevention or treatment of autoimmune disease, comprising the same antibody, and a composition for concurrent administration in combination with a PD-1 antibody or PD-L1 antibody.

Abstract: A cylindrical X-ray tube having an outer insulating layer, a cathode electrode and an anode electrode disposed at both ends of the outer insulating layer, a gate electrode disposed between the cathode and anode electrodes, an emitter, and a target, comprises an inner insulating layer which is disposed between the cathode electrode and the outer insulating layer, is formed to extend downward in a coaxial direction with the outer insulating layer, and is pre-adjusted in order to secure an insulating distance between the cathode electrode and the gate electrode. Thus, by providing a separate internal insulating layer extending coaxially with the external insulating layer between the cathode electrode and the external insulating layer, the insulating distance between the cathode electrode and the gate electrode, the insulating distance between the cathode electrode and the anode electrode may be easily adjusted, so that a desired insulating capability can be secured.

Abstract: A neural network optimizing device includes a performance estimating module that outputs estimated performance according to performing operations of a neural network based on limitation requirements on resources used to perform the operations of the neural network. A portion selecting module receives the estimated performance from the performance estimating module and selects a portion of the neural network which deviates from the limitation requirements. A new neural network generating module generates, through reinforcement learning, a subset by changing a layer structure included in the selected portion of the neural network, determines an optimal layer structure based on the estimated performance provided from the performance estimating module, and changes the selected portion to the optimal layer structure to generate a new neural network. A final neural network output module outputs the new neural network generated by the new neural network generating module as a final neural network.

Abstract: A method of predicting a shape of a semiconductor device includes implementing a modeled semiconductor shape with respect to a designed semiconductor layout, extracting a plurality of samples by independently linearly combining process variables with respect to the modeled semiconductor shape; generating virtual spectrums with respect to ones of the extracted plurality of samples through optical analysis, indexing the virtual spectrums to produce indexed virtual spectrums, generating a shape prediction model by using the indexed virtual spectrums as an input and the modeled semiconductor shape as an output, and indexing a spectrum measured from a manufactured semiconductor device and inputting the spectrum to the shape prediction model to predict a shape of the manufactured semiconductor device.

Abstract: A cylindrical X-ray tube having an outer insulating layer, a cathode electrode and an anode electrode disposed at both ends of the outer insulating layer, a gate electrode disposed between the cathode and anode electrodes, an emitter, and a target, comprises an inner insulating layer which is disposed between the cathode electrode and the outer insulating layer, is formed to extend downward in a coaxial direction with the outer insulating layer, and is pre-adjusted in order to secure an insulating distance between the cathode electrode and the gate electrode. Thus, by providing a separate internal insulating layer extending coaxially with the external insulating layer between the cathode electrode and the external insulating layer, the insulating distance between the cathode electrode and the gate electrode, the insulating distance between the cathode electrode and the anode electrode may be easily adjusted, so that a desired insulating capability can be secured.

Abstract: The present disclosure relates to an antibody specifically binding to epidermal growth factor receptor variant III (EGFRvIII), to nucleic acid coding the antibody, to a vector comprising the nucleic acid, to a host cell, to a preparation method of the antibody, and to a pharmaceutical composition comprising the antibody as an active ingredient for treating a cancer or a tumor.

Abstract: An artificial neural network (ANN) system includes a processor, a virtual overflow detection circuit and a data format controller. The processor performs node operations with respect to a plurality of nodes included in each layer of an ANN to obtain a plurality of result values of the node operations and performs a quantization operation on the obtained plurality of result values based on a k-th fixed-point format for a current quantization of the each layer to obtain a plurality of quantization values. The virtual overflow detection circuit generates a virtual overflow information indicating a distribution of valid bit numbers of the obtained plurality of quantization values. The data format controller determines a (k+1)-th fixed-point format for a next quantization of the each layer based on the generated virtual overflow information. An overflow and/or an underflow are prevented efficiently by controlling the fixed-point format using the virtual overflow.

Abstract: An application producing method regarding a first and second application, the method includes receiving an application generation request related to a second application, the second application being subordinate to a first application, generating the second application and computer readable instructions related to a switch interface for selecting at least one of the first and second applications, the switch interface computer readable instructions related to the switch interface, which when executed by at least one application executing apparatus, causes the application executing apparatus to display the switch interface on at least one of the first and second applications, and when the switch interface is in a first state, the first application is controlled based on a first user input received by the at least one application executing apparatus, and when the switch interface is in a second state, the second application is controlled based on the first user input.

Abstract: A method and apparatus for generating a fixed point neural network are provided. The method includes selecting at least one layer of a neural network as an object layer, wherein the neural network includes a plurality of layers, each of the plurality of layers corresponding to a respective one of plurality of quantization parameters; forming a candidate parameter set including candidate parameter values with respect to a quantization parameter of the plurality of quantization parameters corresponding to the object layer; determining an update parameter value from among the candidate parameter values based on levels of network performance of the neural network, wherein each of the levels of network performance correspond to a respective one of the candidate parameter values; and updating the quantization parameter with respect to the object layer based on the update parameter value.

Abstract: An artificial neural network (ANN) quantization method for generating an output ANN by quantizing an input ANN includes: obtaining second parameters by quantizing first parameters of the input ANN; obtaining a sample distribution from an intermediate ANN in which the obtained second parameters have been applied to the input ANN; and obtaining a fractional length for the sample distribution by quantizing the obtained sample distribution.

Abstract: A method of generating a fixed-point type neural network by quantizing a floating-point type neural network, includes obtaining, by a device, a plurality of post-activation values by applying an activation function to a plurality of activation values that are received from a layer included in the floating-point type neural network, and deriving, by the device, a plurality of statistical characteristics for at least some of the plurality of post-activation values. The method further includes determining, by the device, a step size for the quantizing of the floating-point type neural network, based on the plurality of statistical characteristics, and determining, by the device, a final fraction length for the fixed-point type neural network, based on the step size.

Abstract: A method of quantizing an artificial neural network includes dividing an input distribution of the artificial neural network into a plurality of segments, generating an approximated density function by approximating each of the plurality of segments, calculating at least one quantization error corresponding to at least one step size for quantizing the artificial neural network, based on the approximated density function, and determining a final step size for quantizing the artificial neural network based on the at least one quantization error.