Abstract: Disclosed is a three dimensional mechanical ultrasound probe, which includes a transducer, a driving motor and a transmission mechanism configured to drive the transducer to reciprocatingly oscillate within a predetermined angle. The transmission mechanism includes a transducer base fixed and connected with the transducer, a face gear arranged on the transducer base, and a cylindrical gear meshed with the face gear. A rotating shaft of the face gear is positioned on the transducer base, and the face gear drives the transducer base to reciprocatingly oscillate. A transmission shaft is arranged on the axis of the cylindrical gear, and the transmission shaft is connected with the driving motor. The three dimensional mechanical ultrasound probe provided by the present application aims at simplifying the internal structure of the three dimensional mechanical ultrasound probe, reducing the internal installation complexity and improving motion stability.

Abstract: The present application discloses a method and apparatus for training a semantic representation model, a device and a computer storage medium, which relates to the field of natural language processing technologies in artificial intelligence.

Abstract: The present teaching relates to ranking search content. In one example, a plurality of documents is received to be ranked with respect to a query. Features are extracted from the query and the plurality of documents. The plurality of documents is ranked based on a ranking model and the extracted features. The ranking model is derived to remove one or more documents from the plurality of documents that are less relevant to the query and order remaining documents based on their relevance to the query. The ordered remaining documents are provided as a search result with respect to the query.

Abstract: The present teaching relates to ranking search content. In one example, a plurality of documents is received to be ranked with respect to a query. Features are extracted from the query and the plurality of documents. The plurality of documents is ranked based on a ranking model and the extracted features. The ranking model is derived to remove one or more documents from the plurality of documents that are less relevant to the query and order remaining documents based on their relevance to the query. The ordered remaining documents are provided as a search result with respect to the query.

Abstract: Disclosed is a noise reduction device for heat dissipation fans, the noise reduction device being applied to a heat dissipation fan with an interference structure for rotor and stator blades, wherein the noise reduction device comprises a duct formed at an end portion of a through-flow area of the heat dissipation fan, the inner diameter D of the duct is determined by the size and the rotation speed of the heat dissipation fan, and the axial length L of the duct is determined by an acoustic cutoff condition of the heat dissipation fan.

Abstract: A verification method and a transaction verification method are provided. The verification method applicable for a secure device and an unsecure device utilizes the secure device to generate a match information for a user to see one-to-one substitution relations between a plurality of first verification data and a plurality of second verification data. When the user, after seeing the match information, selects the plurality of second verification data on the unsecure device, the unsecure device transmits a plurality of input data to the secure device, and the secure device converts the plurality of input data into a plurality of to-be-verified passwords according to the match information. The secure device then determines whether or not the plurality of to-be-verified passwords match with a plurality of pre-stored password data in the secure device.

Abstract: In one embodiment, a set of training data consisting of inliers may be obtained. A supervised classification model may be trained using the set of training data to identify outliers. The supervised classification model may be applied to generate an anomaly score for a data point. It may be determined whether the data point is an outlier based, at least in part, upon the anomaly score.

Abstract: Methods and systems are disclosed using linear expressions for machine learning (ML) models for ranking search results. In one example, a method for a computer trains a ML model into a decision tree for ranking search results. The decision tree is converted into a linear expression including Boolean terms. The linear expression is transmitted to one or more search computers that use the linear expression to rank search results for a search query. According to another example, a method for a computer having a search engine includes receiving a linear expression including Boolean terms representing a decision tree. The search engine processes a search query and uses the linear expression to rank search results for the search query.

Abstract: A system and method for providing additional power to a connected load is disclosed. The system includes a power supply unit and an energy storage unit coupled to the power supply unit. A switch has an input coupled to the power supply unit. A power output provides power to the connected load. The power output is coupled to an output of the switch and the energy storage unit. A controller is operative to provide a charging period where the switch is turned on to provide power to the power output. The controller provides a boost period where the switch is controlled via a pulse width modulation signal to connect the power supply with the energy storage unit to supply power to the load when the switch is on, and with the energy storage unit to supply power to the load when the switch is off.

Abstract: A system and method is directed to charging and discharging an energy-tank capacitor of a power supply unit (PSU) in a computing system. The PSU has a forward converter, a buck converter, a switch box, and an option box. The forward converter has a transformer. A primary winding of the transformer is coupled to an input of the PSU, while a secondary winding of the transformer is coupled to an output of the PSU. Two or more components of the forward converter are shared with the buck converter to support the functionality of the buck converter when it is enabled. The switch box has a power switch and a controller switch connected to the forward converter. The energy stored on the energy-tank capacitor is capable of supporting power consumption of the server system during a hold-up time, when an input power to the PSU is interrupted.

Abstract: A system and method is directed to charging and discharging an energy-tank capacitor of a power supply unit (PSU) in a computing system. The PSU has a forward converter, a buck converter, a switch box, and an option box. The forward converter has a transformer. A primary winding of the transformer is coupled to an input of the PSU, while a secondary winding of the transformer is coupled to an output of the PSU. Two or more components of the forward converter are shared with the buck converter to support the functionality of the buck converter when it is enabled. The switch box has a power switch and a controller switch connected to the forward converter. The energy stored on the energy-tank capacitor is capable of supporting power consumption of the server system during a hold-up time, when an input power to the PSU is interrupted.

Abstract: In an embodiment, a method for calculating a noise index value for a digital audio source in a server computer system that is coupled to a plurality of audio sources and configured to operate a teleconference among the plurality of digital audio sources comprises receiving a first digital audio signal from the digital audio source of the one or more digital sources. Using the server computer system, the process identifies two or more types of sounds that are represented in the first digital audio source. The types of sounds identified include at least two of: one or more human voices; a background noise; or an actionable sound that mandates further action. Using a server computer system, calculating the noise index value based upon the types of sounds identified from the first digital audio signal.

Abstract: A novel method and/or system of feature selection is described.

Abstract: In one embodiment, a method includes sharing first content from a first computer of a first participant in an online conference. The first content is shared on a second display of a second computer of a second participant in the online conference. During the sharing of the first content, third content from a third computer of a third participant is queued in a queue at the second computer without display on the second display. For a transition from the first participant to the third participant in the online conference, the third content is shared on the second display of the second computer from the queue.

Abstract: Methods and systems are disclosed using linear expressions for machine learning (ML) models for ranking search results. In one example, a method for a computer trains a ML model into a decision tree for ranking search results. The decision tree is converted into a linear expression including Boolean terms. The linear expression is transmitted to one or more search computers that use the linear expression to rank search results for a search query. According to another example, a method for a computer having a search engine includes receiving a linear expression including Boolean terms representing a decision tree. The search engine processes a search query and uses the linear expression to rank search results for the search query.

Abstract: An example method is provided and includes receiving an input associated with sharing data in a virtual meeting involving a first endpoint and a second endpoint associated with a meeting in a network environment. A window associated with an application that is running on the endpoint is scanned to determine a dataset for sharing, via the virtual meeting, with the second endpoint. The method also includes identify a portion of the dataset to restrict based on analyzing the dataset against a metric. The method also includes preventing a window of the at least one window from being transmitted, via the virtual meeting, to the second endpoint based on a determination that the window includes the portion of the dataset to restrict.

Abstract: Presented herein are techniques to enable a meeting participant/attendee at a first endpoint device in an online/web-based meeting/conference to acquire shared content with high resolution. The first endpoint device receives an initial image of content shared by a second endpoint in the online/web-based meeting/conference. At the first endpoint, the initial image of the shared content is displayed. A user selection of a selected region of the initial image is received at the first endpoint, and a second image of the shared content corresponding to the selected region is received at the first endpoint, wherein the second image has a higher resolution than the initial image.

Abstract: An example method is provided and includes detecting, during a virtual meeting session, a plurality of events associated with each of a plurality of endpoints connected to the virtual meeting session. Each of the endpoints may be associated with a user identifier. The method also includes determining a status value for each of the plurality of endpoints based analyzing one or more of the plurality of events against a metric. The method also includes storing information related to the plurality of events. An output representative of at least one portion of the virtual meeting session is generated based on the stored information. The output identifies a classification of attentiveness associated with a user identifier for the at least one portion of the virtual meeting session.

Abstract: An attendee device in an online conference session views a shared application window in a full screen mode. The attendee device receives data for content being shared by a host device, including at least one application window at a first resolution. The attendee device receives an input to enter a full screen mode for one of the application windows and determines whether displaying that application window at a second resolution for the full screen mode requires the image data to be magnified. When displaying the application window at the second resolution requires magnified image data, the attendee device requests the magnified image data from the meeting server facilitating the conference session. The meeting server provides the magnified image data to the attendee device, and the attendee device displays the application window in full screen mode using the magnified image data.

Abstract: A method is provided in one example and includes establishing a connection with each of a first set of endpoints associated with meeting attendees, and establishing a first online meeting between the first set of endpoints associated with meeting attendees in which the first online meeting has a meeting identifier associated therewith. The method further includes receiving first context information associated with the first online meeting from at least one of the first set of endpoints associated with meeting attendees, and establishing a second online meeting between a second set of endpoints associated with meeting attendees without disconnecting at least one of the first set of endpoints associated with meeting attendees in which the second online meeting has the same meeting identifier as the first online meeting associated therewith. The method still further includes maintaining the first context information within the second online meeting.