Abstract: A cloud platform includes several web services that facilitate the automated tuning and deployment of pre-trained deep learning models configured for software engineering tasks. The automated tuning and deployment allow a developer to fine-tune a pre-existing model without having access to the parameters of the pre-existing and the fine-tuned model in a manner that does not require user management input. The cloud platform provides a set of files for each pre-trained models used to automatically build a fine-tuning infrastructure to fine-tune a model and a deployment infrastructure that deploys the fine-tuned model without requiring user input.

Abstract: A phase locked loop having a charge pump is described. The charge pump relies on close matching of FETs (Field Effect Transistor) electrical parameters to FETs in a current reference circuit. To achieve close matching of FET electrical performance, FEOL (Front End Of Line), comprising all FET shapes, of the current pump is identical in shapes and layout to the current reference circuit. BEOL (Back End Of Line) differs between the charge pump and the current reference circuit. The charge pump and the current reference circuit are arranged in a row. A shield circuit having FEOL shapes and layout identical to the current pump may be placed at each end of the row.

Abstract: The present invention provides a method to establish cloned game accounts, when an auction information is posted on the game account trading platform, the game account trading platform will establish a testing game interface according to the game interface under the relevant game account, and generate a cloned game account based on the game objects, so that the buyer user primary account can enter this testing game interface to test the cloned game account.

Abstract: A method includes obtaining feature generation code from, which is configured to determine features relating to input data. The method further includes obtaining data grouping code, which is configured to generate training data by determining a plurality of data groupings for the features relating to the input data. The method further includes obtaining modeling code, which is derived at least in part by applying one or more machine learning algorithms to the training data. The method further includes applying a model wrapper code to the feature generation code, the data grouping code, and the modeling code to generate a model wrapper and deploying the model wrapper such that the model wrapper may receive a first application programming interface (API) call including an input data value, determine a score relating to the input data value, and send a second API call including the score in response to the first API call.

Abstract: The present invention is publish a method and system for game account valuation, through the establishment of a game account trading platform, provide different game accounts for valuation, acquisition and consignment, seller users can choose a suitable plan according to their personal needs, and through the management of the game account trading platform, increase the credibility and security of transaction information, and also can reduce the impact on the value of game data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing data generated by a sensing system that rotationally senses an environment. In one aspect, a method comprises partitioning a predetermined period of time into a plurality of sub-periods, wherein the predetermined period of time is a period of time for which data generated by the sensing system constitutes a complete rotational sensing of the environment; for each sub-period: receiving current data generated by the sensing system during the sub-period and characterizing a respective partial scene of the environment; processing the current data using an object detection neural network to generate a current object detection output that is specific to the respective partial scene of the environment.

Abstract: A doorbell system includes a power supply circuit, a smart module, a chime, a chime switch and a control module. The smart module provides a smart function. The chime switch is electrically connected between the control module, the smart module, and the power supply circuit. When the chime switch is turned on, the current flows through the chime under control of the control module. When the chime switch is turned off, the current cannot flow through the chime under control of the control module; consequently, the buzzing sound cannot be generated.

Abstract: The present invention relates to compositions and methods for selectively etching silicon nitride in the presence of silicon oxide, polysilicon and/or metal silicides at a high etch rate and with high selectivity. Additives are described that can be used at various dissolved silica loading windows to provide and maintain the high selective etch rate and selectivity.

Abstract: A lithography method to pattern a first semiconductor wafer is disclosed. An optical mask is positioned over the first semiconductor wafer. A first region of the first semiconductor wafer is patterned by directing light from a light source through transparent regions of the optical mask. A second region of the first semiconductor wafer is patterned by directing energy from an energy source to the second region, wherein the patterning of the second region comprises direct-beam writing.

Abstract: An electronic device, a method, and a computer-readable recording medium are provided for deactivating a chat room of a messenger application. The electronic device is configured for: receiving a deactivation request for one or more chat rooms, in which the deactivation request is associated with a chat room configuration, and the chat room configuration includes a deactivation condition for the one or more chat rooms; storing the chat room configuration including the deactivation condition for the one or more chat rooms based on the deactivation request; and deactivating the one or more chat rooms based on determining whether or not the deactivation condition for the one or more chat rooms is satisfied.

Abstract: A signal processing device includes an echo estimation device, a captured signal buffer device and a delay estimation device. The echo estimation device generates an echo estimation signal according to a reference signal and a set of reflection path simulation coefficients and compensates the echo estimation signal according to a first delay to generate a compensated echo estimation signal. The captured signal buffer device buffers a captured signal captured by microphone device and outputs the captured signal according to a second delay to generate a compensated captured signal. The delay estimation device estimates an amount of delay adjustment according to the compensated echo estimation signal and the compensated captured signal and updates the first delay or the second delay according to the amount of delay adjustment. A difference between an upper bound and a lower bound of the first delay is smaller than or equal to 1.

Abstract: A method for improving magnetic properties of a Ce—Y-rich rare earth permanent magnet is provided, and the Ce—Y-rich rare earth permanent magnet is subjected to pressurized heat treatment to improve magnetic properties. The method includes: preparing a pristine magnet through a sintering process; and placing the pristine magnet into a pressurized heat treatment device and performing pressurized heat treatment under the protection of an argon atmosphere. By regulating parameters such as pressure, temperature and holding time in the heat treatment process, element diffusion in the Ce—Y-rich permanent magnet is promoted, and coercivity, remanence, magnetic energy product and temperature stability of the Ce—Y-rich permanent magnet are improved. The method has advantages of a simple process with low energy consumption, a substitution amount of rare earths Ce—Y up to 90 wt % while having excellent magnetic performance, so that a way for efficient utilization of high-abundance rare earths Ce and Y is provided.

Abstract: Aspects of the disclosure provide for automatically generating labels for sensor data. For instance, first sensor data for a vehicle may be identified. This first sensor data may have been captured by a first sensor of the vehicle at a first location during a first point in time and may be associated with a first label for an object. Second sensor data for the vehicle may be identified. The second sensor data may have been captured by a second sensor of the vehicle at a second location at a second point in time outside of the first point in time. The second location is different from the first location. A determination may be made as to whether the object is a static object. Based on the determination that the object is a static object, the first label may be used to automatically generate a second label for the second sensor data.

Abstract: An indoor positioning system, for detecting a position of a target object located in a containing body, includes a tracking label, a plurality of positioning label groups, and a scanning device. The tracking label is disposed on the target object. Each positioning label groups is disposed on the containing body and has a plurality of label units. Each label unit defines a positioning interface, which forms a plurality of areas of the containing body. The scanning device senses and reads the tracking label and each label unit of each positioning label group. The scanning device obtains the positioning interface in which the tracking label is located through an analysis computation according to a signal relationship between the tracking label and each label unit, thereby identifying the positioning interface to obtain one of the areas of the target object in the containing body.

Abstract: An automated system for translating source code written in one programming language into a different programming language utilizes a neural transformer with attention trained on semi-supervised data. The model is jointly pre-trained with a masked language model objective and an autoregressive objective on a large unsupervised source code corpus to learn to comprehend the syntactic structure and semantics of source code. The pre-trained model is then fine-tuned with a token-type prediction objective and an autoregressive objective on supervised translation tasks and data augmented tasks to learn to translate source code from one programming language into a different programming language.

Abstract: A channel estimation method is configured to estimate a channel length. The method includes the following operations: receiving an input signal; summing the input signal and an analog echo cancelation signal decrease an echo of the input signal, and generate a first signal according to a result of the summation; providing an analog gain value to the first signal to generate a second signal; performing an analog-to-digital conversion to the second signal to generate a third signal; obtaining a ratio according to an energy of a first frequency and an energy of a second frequency of the third signal; and estimating the channel length according to the ratio, and setting the analog gain value according to the estimated channel length.

Abstract: A method of forming a semiconductor structure includes forming first semiconductor devices over a first substrate, forming a first dielectric material layer over the first semiconductor devices, forming vertical recesses in the first dielectric material layer, such that each of the vertical recesses vertically extends from a topmost surface of the first dielectric material layer toward the first substrate, forming silicon nitride material portions in each of the vertical recesses; and locally irradiating a second subset of the silicon nitride material portions with a laser beam.

Abstract: A channel estimation method is configured to estimate a channel length. The method includes the following operations: receiving an input signal; summing the input signal and an analog echo cancelation signal decrease an echo of the input signal, and generate a first signal according to a result of the summation; providing an analog gain value to the first signal to generate a second signal; performing an analog-to-digital conversion to the second signal to generate a third signal; obtaining a ratio according to an energy of a first frequency and an energy of a second frequency of the third signal; and estimating the channel length according to the ratio, and setting the analog gain value according to the estimated channel length.

Abstract: Embodiments of the invention are directed to using a trained machine learning model to generate predicted circuit data for a circuit design, and computing an objective function using the predicted circuit data. Optimization of the objective function is performed to generate an optimal solution, and the optimal solution is mapped to the circuit design.

Abstract: A noise cancellation circuit includes: a first filter circuit for filtering a first input signal according to a first filter coefficient to generate a first filtered signal; a signal processing circuit for generating a feedback signal according to a second input signal and an audio signal; a second filter circuit for filtering the feedback signal according to a second filter coefficient to generate a second filtered signal; a first multiplication circuit for multiplying the first filtered signal by a first scale to generate a first intermediate signal; a second multiplication circuit for multiplying the second filtered signal by a second scale to generate a second intermediate signal; a first adder circuit for adding the first intermediate signal to the second intermediate signal to generate a noise cancellation signal; and a second adder circuit for adding the noise cancellation signal to the audio signal to generate an output signal.