Abstract: The present invention relates to the field of catalytic cracking, and discloses a catalyst with regular structure capable of simultaneously reducing the emission of SOx and NOx and a preparation method thereof, and a method for simultaneously reducing both SOx and NOx from flue gas, the catalyst comprises a support with regular structure and an active component coating distributed on the inner surface and/or the outer surface of the support with regular structure, the active metal component contains: 1) as oxide, 50-95 wt % of metal component(s) selected from Group rare earth and/or Group IIA; 2) as oxide, 5-50 wt % of non-precious metal component(s) selected from Groups VB, VIIB, VIII, IB, and IIB; 3) as element, 0.01-2 wt % of precious metal component. Using the catalyst provided by the invention can reduce the total amount of added active components and enhance the emission reduction effect of the additive.

Abstract: A system is described for redirecting multimedia in a collaborative session on a virtual desktop. The virtual desktop session can be established, and collaborator virtual desktop clients can be connected in a collaborative session where each collaborator can view the desktop GUI in their respective virtual desktop client. A request can be received to play media in a media player in the virtual desktop. The media stream can be intercepted in the virtual desktop before it is rendered in the media player and conveyed to each collaborator's client over a separate virtual channel established between the virtual desktop and each collaborator. The data stream can then be rendered in a client media player by each collaborator's client.

Abstract: A catalyst for simultaneously reducing both SOx and NOx in flue gas and a preparation method and use thereof are provided. They catalyst contains a support or inorganic oxide matrix, a rare earth metal, a non-precious metal selected from Group VIII, or non-precious metal(s) selected from Groups VB, VIII, IB, and IIB, a precious metal, an optional Group VIIB non-precious metal, and an optional Group IIA metal. Contacting the flue gas with the catalyst simultaneously reduces both SOx and NOx in the flue gas.

Abstract: The present disclosure relates to training method and apparatus for an object recognition model. There provides a training sample optimization apparatus for a neural network model for object recognition, comprising: for each training sample in a training sample database, a fluctuation determination unit configured to determine a fluctuation of model prediction of the training sample relative to a corresponding labeled identity of the training sample in a case of training the neural network model; and an optimization unit configured to determine whether the training sample can be used for training of the neural network model in the next training epoch, based on the fluctuation of the training sample.

Abstract: The present invention relates to method and apparatus for training a neural network for object recognition. A training method which includes inputting a training image set containing an object to be recognized, dividing the image samples in the training image set into simple samples and hard samples, for each kind of the image sample and the variation image sample, performing, a transitive transfer, calculating a distillation loss of the transferred student feature of the image sample relative to a teacher feature extracted from corresponding image sample of the other kind, classifying, the image sample, and calculating a classification loss of the image sample, calculating a total loss related to the training image set, and updating parameters of the neural network according to the calculated total loss.

Abstract: A method and device for testing a positioning and speed measuring system main unit. The method includes: obtaining a first simulation signal by simulating an absolute position sensor and a relative position sensor; sending the first simulation signal to a main unit of a to-be-tested positioning and speed measuring system; obtaining a first result calculated by the main unit based on the first simulation signal; obtaining a second result calculated based on the first simulation signal, the second result being a reference result corresponding to the first simulation signal; and determining that the main unit is abnormal if a comparison between the first result and the second result exceeds a first preset range. Therefore, the main unit can still be tested without components including absolute position sensors and relative position sensors, the test process is simple and easy to implement, a test environment can be built without large-scale investment.

Abstract: Systems and methods are provided for dynamically optimizing and configuring various aspects of virtual desktops in virtual desktop infrastructure. Data collectors can be installed on and operate on various components in the virtual desktop infrastructure, such as on the virtual desktops running on the server, on the virtual desktop clients running on user devices, and on the connection server. The data collectors can operate to collect various types of information from corresponding components, such as application usage data and status, device performance, networking environment and speed, application or system crash data, and so on. The collected data can be logged, tracked, and analyzed to perform various actions on the virtual desktop.

Abstract: A system is described for redirecting multimedia in a collaborative session on a virtual desktop. The virtual desktop session can be established, and collaborator virtual desktop clients can be connected in a collaborative session where each collaborator can view the desktop GUI in their respective virtual desktop client. A request can be received to play media in a media player in the virtual desktop. The media stream can be intercepted in the virtual desktop before it is rendered in the media player and conveyed to each collaborator's client over a separate virtual channel established between the virtual desktop and each collaborator. The data stream can then be rendered in a client media player by each collaborator's client.

Abstract: The present disclosure provides training and application methods and apparatuses for a neural network model, and a storage medium. The training method includes: quantizing, in a forward transfer process, a network parameter represented by a continuous real value, and calculating a quantization error; determining, in a backward transfer process, a gradient of a weight in the neural network model; correcting the gradient of the weight based on the calculated quantization error, wherein the correcting includes correcting a magnitude of the gradient and correcting a direction of the gradient; and updating the neural network model according to the corrected gradient.

Abstract: The present disclosure relates to training method and apparatus for an object recognition model. There provides a training sample optimization apparatus for a neural network model for object recognition, comprising: for each training sample in a training sample database, a fluctuation determination unit configured to determine a fluctuation of model prediction of the training sample relative to a corresponding labeled identity of the training sample in a case of training the neural network model; and an optimization unit configured to determine whether the training sample can be used for training of the neural network model in the next training epoch, based on the fluctuation of the training sample.

Abstract: The present invention relates to method and apparatus for training a neural network for object recognition. A training method which includes inputting a training image set containing an object to be recognized, dividing the image samples in the training image set into simple samples and hard samples, for each kind of the image sample and the variation image sample, performing, a transitive transfer, calculating a distillation loss of the transferred student feature of the image sample relative to a teacher feature extracted from corresponding image sample of the other kind, classifying, the image sample, and calculating a classification loss of the image sample, calculating a total loss related to the training image set, and updating parameters of the neural network according to the calculated total loss.

Abstract: A training method and device for an object recognition model. An apparatus for optimizing a neural network model for object recognition, including a loss determination unit configured to determine loss data for features extracted from a training image set using the neural network model and a loss function with a weight function, and an updating unit configured to perform an updating operation on parameters of the neural network model based on the loss data and an updating function, wherein the updating function is derived based on the loss function with the weight function of the neural network model, and the weight function and the loss function change monotonically in a specific value interval in the same direction.

Abstract: One of the aspects of the present invention discloses a feature point detection method. The method comprises: acquiring a face region in an input image; acquiring first positions of first feature points and second feature points according to a pre-generated first model; estimating second positions of the first feature points according to the first positions of the first feature points and pre-generated second models; detecting third positions of the first feature points and the second feature points according to the second positions of the first feature points, the first positions of the second feature points and pre-generated third models. According to the present invention, the final detected face shape could approach to the actual face shape much more.

Abstract: One of the aspects of the present invention discloses a feature point detection method. The method comprises: acquiring a face region in an input image; acquiring first positions of first feature points and second feature points according to a pre-generated first model; estimating second positions of the first feature points according to the first positions of the first feature points and pre-generated second models; detecting third positions of the first feature points and the second feature points according to the second positions of the first feature points, the first positions of the second feature points and pre-generated third models. According to the present invention, the final detected face shape could approach to the actual face shape much more.