Abstract: A method includes receiving a sequence of acoustic frames as input to an automatic speech recognition (ASR) model. The method also includes generating, by a first encoder, a first higher order feature representation for a corresponding acoustic frame. The method also includes generating, by a second encoder, a second higher order feature representation for a corresponding first higher order feature representation. The method also includes generating, by a language identification (ID) predictor, a language prediction representation based on a concatenation of the first higher order feature representation and the second higher order feature representation. The method also includes generating, by a first decoder, a first probability distribution over possible speech recognition hypotheses based on a concatenation of the second higher order feature representation and the language prediction representation.

Abstract: A device for transmitting commands with a circuit of a circuit board to test a connection interface, a system and a method thereof are disclosed. In the system, an inter-integrated circuit (I2C) of the circuit board under test is used to transmit an control command to a test device, which is connected to the inter-integrated circuit via a memory connection interface of the circuit board under test, the test device converts the control command to test the memory connection interface connected thereto, so as to achieve the technical effect of improving test efficiency in testing the memory connection interface of the circuit board under test without using an external connection line.

Abstract: The present invention relates to an amphiphilic material and an application thereof in preparation of a liposome, and in particular, the present invention provides an amphiphilic material, wherein the structure of the amphiphilic material is as follows. The amphiphilic material of the present invention is used for preparing drug-loaded nanoparticles to effectively enter cells such as tumor cells, thereby enhancing the therapeutic effect of the drug.

Abstract: Disclosed herein are synthesis and use of covalent inhibitors selective for Transcriptional Enhancer Factor TEF-1 (TEAD1), which can be used for treatment of cancers such as glioblastoma, gastric cancer, colorectal cancer, pancreatic ductal adenocarcinoma (PDAC), and malignant pleural mesothelioma (MPM). Further disclosed herein are pharmaceutical compositions including the TEAD1 inhibitor and methods of treating cancers using the same.

Abstract: Compounds active on protein kinases are described, as well as methods of making and using such compounds to treat diseases and conditions associated with aberrant activity of protein kinases.

Abstract: A cloud-based platform may receive an indication of one or more event associated with a service provider and modify a configuration associated with providing a schedule of respective service appointments offered by the service provider based on the indication of the one or more events. The cloud-based platform may then receive a request for a new service appointment after modifying the configuration and in response to receiving the request for the new service appointment, determine a number of scheduled appointments associated with a first time window, determine whether the number of scheduled appointments is less than a maximum number of appointments associated with the first time window, and automatically schedule the new service appointment during the first time window in response to determining that the number of scheduled appointments is less than the maximum number of appointments associated with the first window.

Abstract: Semiconductor structures and methods for forming the same are provided. In one form, a semiconductor structure includes: a substrate; a bottom dielectric layer, located on the substrate; a bottom interconnect layer, located in the bottom dielectric layer; a top dielectric layer, located on the bottom dielectric layer and the bottom interconnect layer; a conductive plug, located in the top dielectric layer on a top of the bottom interconnect layer and having a bottom in direct contact with the bottom interconnect layer and a sidewall in direct contact with the top dielectric layer; a top interconnect layer, located in the top dielectric layer above the conductive plug and in contact with the conductive plug; and a top adhesion layer, located between the top interconnect layer and the top dielectric layer. By means of embodiments and implementations of the present disclosure, electrical connection performance of a semiconductor structure is optimized.

Abstract: In certain aspects, a memory device includes a memory cell array having rows of memory cells, word lines respectively coupled to the rows of memory cells, and a peripheral circuit coupled to the memory cell array through the word lines. Each memory cell is configured to store a piece of N-bits data in one of 2N levels, where N is an integer greater than 1. The level corresponds to one of 2N pieces of N-bits data. The peripheral circuit is configured to program, in a first pass, a row of target memory cells, such that each of the row of target memory cells is programmed into one of 2N/m intermediate levels based on the piece of N-bits data to be stored in the target memory cell, where m is an integer greater than 1. The peripheral circuit is also configured to program, in a second pass after the first pass, the row of target memory cells, such that each target memory cell is programmed into one of the 2N levels based on the piece of N-bits data to be stored in the target memory cell.

Abstract: A computer implemented method for pose estimation of an image sensor includes receiving an omnidirectional image of a scene captured by an image sensor; using a trained neural network to generate a rotation equivariant feature map from the omnidirectional image of the scene; and determining information relating to the pose of said camera when capturing the scene from the rotation equivariant feature map. The rotation equivariant feature map is a SO(3) indexed feature map.

Abstract: The invention relates to A method for operating a fuel-operated burner (3) downstream of an engine (1) and upstream of a catalytic converter (4), comprising an actuation of an ignition apparatus (12) of the burner (3) during a pre-definable preheating phase, without delivering fuel (11) to the burner (3) during the preheating phase, and, after the end of the preheating phase, a timely delivery of fuel (11) to the burner (3) and combustion of the delivered fuel (11) in the burner (3). Further, a computing unit and a computer program product for carrying out such a method are proposed.

Abstract: A method (200) for diagnosing the functionality of a burner (100), comprising: determining a lambda value (210) of an exhaust gas of the burner (100), comparing (220) the determined lambda value to a time-based changing lambda threshold (225), and detecting a malfunction (280) when the determined lambda value exceeds the time-based changing lambda threshold (225). Furthermore, a computing unit and a computer program for carrying out such a method (200) are proposed.

Abstract: Method and apparatus of processing a sequence of video frames comprising generating at least one video frame and using an analog neural network to select, within the at least one video frame, at least one patch of pixels and process the at least one pixel patch to produce a patch feature for each of the at least one pixel patches. The method digitizes the patch feature, identifies objects within the digitized patch feature, and tracks the objects to generate control information that is used by the analog neural network to select and process the pixel patches.

Abstract: A liveness test method and apparatus is disclosed. The liveness test method includes detecting a face region in an input image for a test target, implementing a first liveness test to determine a first liveness value based on a first image corresponding to the detected face region, implementing a second liveness test to determine a second liveness value based on a second image corresponding to a partial face region of the detected face region, implementing a third liveness test to determine a third liveness value based on an entirety of the input image or a full region of the input image that includes the detected face region and a region beyond the detected face region, and determining a result of the liveness test based on the first liveness value, the second liveness value, and the third liveness value.

Abstract: The present application relates to the technical field of grinding, and in particular, to a small grinder, which solves the problems of complex structure, large volume and inconvenience in carrying in the prior art. A small grinder provided by the present application, comprising a housing; a fixing plate is provided inside the housing, the fixing plate divides an inner portion of the housing into an upper portion and a lower portion, and a feeding assembly is provided on the upper portion of the housing; a bracket is further provided on the upper portion of the housing; the bracket divides the upper portion of the housing into two parts, and the two parts are respectively provided with a power mechanism and a power supply mechanism; and the male grinding head is rotatably fitted in the female grinding head. The small grinder has a small volume, a reasonable layout, a stable structure, and simple and convenient operation, and is easy to maintain and carry.

Abstract: A shift register, a gate drive circuit, and a display panel are provided. The shift register includes an input sub-circuit configured to pre-charge a pull-up node using an input signal; an output sub-circuit configured to output a clock signal through an signal output terminal; a pull-down control sub-circuit configured to control a potential of a pull-down node using a power supply voltage signal; a first pull-down sub-circuit configured to pull down a potential of the pull-down node using a first preset voltage signal; and a first control sub-circuit configured to control the potential of the pull-up node using a second preset voltage signal in response to the potential of the pull-down node; a potential of the first preset voltage signal is lower than a potential of a non-operating level signal of the first pull-down sub-circuit, but higher than a potential of the second preset voltage signal.

Abstract: A control system for an autonomous vehicle or robot comprises a plurality of high level controllers. Each high level controller is able to provide high level movement commands independently of the other high level controllers. A low level controller is arranged to receive the high level movement commands of one of the high level controllers and to convert said received high level movement commands into electrical outputs to a plurality of electrical motors/actuators for driving the vehicle/robot. A decision system, independent of the high level controllers, is configured to decide which one of the high level controllers is to be active. The active high level controller only is used provide the high level movement commands to the low level controller.

Abstract: An image processing apparatus and method are provided. The image processing apparatus acquires a target image including a depth image of a scene, determines three-dimensional (3D) point cloud data corresponding to the depth image based on the depth image, and extracts an object included in the scene to acquire an object extraction result based on the 3D point cloud data.

Abstract: Techniques are disclosed for storing an arranging data in a database. A method includes a computer system storing, in a database, data indicative of a graph data structure having a plurality of nodes connected by a plurality of edges. The method further includes the computer system determining that a number of edges connected to a first node satisfies a threshold number. In response to the determining, the computer system may store an index in an index row associated with the first node. The index identifies a first row having first and second ranges of values stored in first and second rows, respectively. The values in the first and second rows correspond to edges connected to the first node. The values in the first and second ranges are usable to indicate properties of corresponding ones of the plurality of edges.

Abstract: A nondestructive testing method for detecting and distinguishing internal and external defects of a wire rope includes: acquiring a magnetic flux signal and a MFL signal of a detected wire rope; preprocessing the magnetic flux signal and the MFL signal of the detected wire rope; comparing a preprocessed magnetic flux signal and a preprocessed MFL signal with a preset magnetic flux signal threshold and a preset MFL signal threshold respectively, and calculating a defect position; extracting a magnetic flux signal of a defect and an MFL signal of the defect based on the defect position; calculating a defect width of the detected wire rope based on the magnetic flux signal of the defect and the MFL signal of the defect; calculating a defect cross-sectional area loss of the detected wire rope based on the defect width; and determining whether the defect is the internal defect or the external defect.