Abstract: The present disclosure provides a method and apparatus for scheduling a plurality of available graphics processing units (GPUs). Multiple GPU pools may be set, wherein each GPU pool is configured to serve one or more jobs requiring the same number of GPUs. Available GPUs may be assigned to each GPU pool. A job and job information related to the job may be received, wherein the job information indicates a number of GPUs required for performing the job. A corresponding GPU pool may be selected from the multiple GPU pools based at least on the job information. Available GPUs to be scheduled to the job in the selected GPU pool may be determined based at least on the job information. In addition, the determined available GPUs may be scheduled to the job.

Abstract: A display system includes an integrated laser and modulator device and a display assembly. The integrated laser and modulator device includes a laser component configured to facilitate light emission responsive to applied current and a modulator component configured to selectively modulate light responsive to applied signal. The modulator component is integrally coupled to the laser component via a bridging structure that intervenes between the laser component and the modulator component. At least a portion of the bridging structure facilitates power reflectivity into a laser cavity of the laser component. The bridging structure facilitates transmission of light emitted by the laser component into the modulator component for modulation by the modulator component to provide modulated light. The display assembly is configured to direct the modulated light provided by the integrated laser and modulator device to illuminate pixels to form an image.

Abstract: The techniques disclosed herein provide dynamic curation of sequence events for communication sessions. A system can utilize smart filtering techniques to generate and select sequence events that are designed to optimize user engagement. The system can collect contextual data associated with a communication session, which can be in the form of a private chat session, a multi-user editing session, a group meeting, a live broadcast, etc. The system can utilize the contextual data, and other input data defining user activity, to customize sequence events defining contextually-relevant user interface (UI) layouts, volume levels, camera angles, special effects, and other parameters controlling aspects of the communication session.

Abstract: The disclosure herein describes a system for calculating meeting inclusion metrics including insights and recommendations. Meeting data associated with one or more meetings attended by at least one participant remotely is converted into anonymized meeting data for inclusivity metric analysis. An inclusivity insights manager generates inclusivity metrics associated with inclusive behavior and language occurring during meetings to measure the level of inclusivity. The inclusivity metrics include attendee participation metrics measuring an amount of participation by each meeting attendee, participation in-person versus participation remotely, concurrent speech indicating attendees may be talking over one another or other interruptions occurring during meetings. Inclusivity metric data includes insights and actionable recommendations to improve inclusivity at future meetings provided at an individual level, group level or organizational level.

Abstract: The techniques disclosed herein provide an enhanced single sign-on flow for secure computing resources, such as a virtual machine or hosted applications. In some configurations, the techniques process different types of security data, e.g., credentials, tokens, certificates, and reference objects at specific computing entities of a system to provide a single sign-on flow for providing access to secure computing resources from a client computing device. In one illustrative example, a select type of security data, such as a certificate, is generated from a token and a claim at a particular computing resource, such as an agent operating on a virtual machine. In another example, a signed version of the certificate can be stored and verified at the virtual machine. By generating certificates at such particular computing resources, the computing resource can verify a person's credentials using a secure single sign-on flow without requiring the person to provide credentials multiple times.

Abstract: Compressing memory addresses within an execution trace via reference to a translation lookaside buffer (TLB) entry. A microprocessor identifies a TLB entry within a TLB slot, the TLB entry mapping a virtual memory page to a physical memory page. The microprocessor initiates logging of the TLB entry by initiating logging of at least a virtual address of the virtual memory page, and an identifier that uniquely identifies the TLB entry from among a plurality of live TLB entries. Subsequently, the microprocessor identifies a cache entry within a memory cache slot, the cache entry comprising a physical memory address corresponding to a cache line. The microprocessor initiates logging of the cache entry by matching a physical memory page identification portion of the physical memory address with the TLB entry, and then initiates logging of at least the identifier for the TLB entry and an offset portion.

Abstract: An HMD includes a single photon avalanche diode (SPAD) array comprising a plurality of SPAD pixels. The HMD also includes a display positioned to display images for viewing by an eye of a user. The HMD also includes one or more processors and one or more hardware storage devices storing instructions that are executable by the one or more processors to configure the HMD to perform various acts associated with using the SPAD array to capture an image frame of an environment for display to the user.

Abstract: The present disclosure provides method and apparatus for automatically generating motions of an avatar. A message in a session between a user and an electronic conversational agent may be obtained, the avatar being a visual representation of the electronic conversational agent. At least one facial animation and/or body animation may be determined based on at least one part of the message. At least one motion of the avatar may be generated based at least on the facial animation and/or the body animation.

Abstract: A method of performing a write operation in a holographic data storage system, in which schedule schedules at least one write operation across multiple non-contiguous write intervals, the write operation pertaining to a set of data to be stored in a region of a holographic recording medium. In each of the non-contiguous write intervals, the region of the holographic recording medium is exposed to an interference pattern caused by interference between a reference beam and an input beam carrying the set of data. The multiple non-contiguous write intervals have a total aggregate duration of sufficient length to cause a persistent state change in the exposed region, such that the set of data is recoverable from that region by the end of a final write interval of the multiple non-contiguous write intervals.

Abstract: Methods, systems, and computer storage media for providing a tailored meeting-video segment associated with a meeting-video management engine of a meeting-video management system. The tailored meeting-video segment corresponds to a portion of meeting-video content that is programmatically generated based on features associated with video data, meeting data, and user data. A tailored meeting-video segment—or a plurality of tailored meeting-video segments—can be generated by employing a meeting-video tailoring machine learning model of the meeting-video management engine. In particular, the features—associated with video data comprising the plurality of clips, meeting data of the meeting, and user data of the user—are meeting-video tailoring features used by the meeting-video tailoring machine learning model to generate the tailored meeting-video segment.

Abstract: Examples are disclosed related to optically transparent antennas. One example provides a device, comprising an electrically insulating substrate that is at least partially optically transparent, one or more antennas disposed on the electrically insulating substrate, each antenna comprising a film of a conductive material that is at least partially optically transparent, the one or more antennas comprising a communication antenna, and processing circuitry electrically coupled to the communication antenna, the processing circuitry configured to one or more of send or receive signals via the communication antenna.

Abstract: An enterprise system server, a computer-readable storage medium, and a method for targeted training of inductive multi-organization recommendation models for enterprise applications are described herein. The method includes receiving enterprise application data from remote organization computing systems executing the enterprise application, training per-organization recommendation models for a subset of the organizations, and validating each per-organization recommendation model on enterprise application data corresponding to one or more other organizations.

Abstract: A system and method for converting a document is described. The system accesses a document comprising one or more section breaks. The system detects sections of the text document demarked by the one or more section breaks and generates a section title metadata and a section summary metadata for each section of the plurality of sections. The system inserts the section title metadata and the section summary metadata at the corresponding section breaks in the text document. The system modifies the text document into slides. Each slide being formed for each section based on the corresponding section title metadata and the section summary metadata. The system generates a presentation document based on the slides.

Abstract: A computing system that enables data stored in a persistent memory region to be preserved when a processor fails can include volatile memory comprising the persistent memory region, non-volatile memory, and a system on a chip (SoC). The SoC can include a main processor that is communicatively coupled to both the volatile memory and the non-volatile memory. The SoC can also include an auxiliary processor that is communicatively coupled to both the volatile memory and the non-volatile memory. The SoC can also include instructions that are executable by the auxiliary processor to cause the data in the persistent memory region of the volatile memory to be transferred to the non-volatile memory in response to a failure of the main processor.

Abstract: Computerized systems are provided for automatically determining action items of an event, such as a meeting. The determined action items may be personalized to a particular user, such as a meeting attendee, and may include contextual information enabling the user to understand the action item. In particular, a personalized action item may be determined based in part from determining and utilizing particular factors in combination with an event dialog, such as an event speaker's language style; user role in an organization; historical patterns in communication; event purpose, name, or location; event participants, or other contextual information. Particular statements are evaluated to determine whether the statement likely is or is not an action item. Contextual information may be determined for action items, which then may be provided to the particular user during or following the event.

Abstract: Embodiments of the present disclosure provide systems, methods, and non-transitory computer storage media for detecting abnormal behavior of device in an enterprise network based on an analysis of behavioral information of the device's neighbors in network. At a high level, embodiments of the present disclosure employ a hive-mind approach to determine anomalous behavior of a device in a network based on analyzing behavior information reported by neighboring devices within the network. Embodiments identify that a device is alive and connected within the network based on multiple neighboring devices reporting behavioral information about the device; however, the device may be dysfunctional and failing to report its own information. By aggregating and analyzing behavioral information of a device based on the reporting information of its neighboring devices, embodiments of the present disclosure are able to determine whether a device is healthy even when the device is unable to report its own information.

Abstract: The automated generation of a web site preview based on a web site identifier, where the web site preview has one or more controls that facilitate connection to one or more endpoints. In response to receiving the web site identifier from the client computing system, the service uses the web site identifier to navigate to the web site identified by the web site identifier. After navigating to that web site, the service accesses an action identifier and an associated endpoint identifier from the web site. The service then generates preview data representing a preview of the web site represented by the web site identifier. The service provides the preview data to the client, where the preview with the control is rendered, so as to facilitate connection with the associated endpoint of the control.

Abstract: A neural transformer model with attention is trained to predict candidates to complete a line of source code with a zero-inference capability. The model is trained on an unsupervised training dataset that includes features from source code written in multiple programming languages. The features include a file-level context and a local context, where the file-level context includes a global context, a class context, a function context, and/or a method context for each class, function and/or method of the source code programs used in the training dataset. The local context includes method bodies, function bodies, and/or stand-alone code of main method routines. From these features, the model is able to learn to predict an ordered sequence of code elements that complete a line of source code in a programming language seen and not seen during training.

Abstract: Innovations in unified intra block copy (“BC”) and inter prediction modes are presented. In some example implementations, bitstream syntax, semantics of syntax elements and many coding/decoding processes for inter prediction mode are reused or slightly modified to enable intra BC prediction for blocks of a frame. For example, to provide intra BC prediction for a current block of a current picture, a motion compensation process applies a motion vector that indicates a displacement within the current picture, with the current picture being used as a reference picture for the motion compensation process. With this unification of syntax, semantics and coding/decoding processes, various coding/decoding tools designed for inter prediction mode, such as advanced motion vector prediction, merge mode and skip mode, can also be applied when intra BC prediction is used, which simplifies implementation of intra BC prediction.

Abstract: Techniques are described herein that are capable of constructing a finite automaton using regular expression derivatives to simulate behavior of a backtracking engine. The behavior indicates an order in which paths in an input regular expression are to be evaluated by the backtracking engine. The finite automaton is constructed to include a graph that includes a root node that represents the input regular expression. Regular expressions are derived such that each derived regular expression is a regular expression derivative of the input regular expression or of another derived regular expression. Priorities are assigned to alternations in the derived regular expressions to correspond to priorities indicated by the behavior. Nodes that represent the respective derived regular expressions and transitions between respective pairs of nodes are caused to be included in the graph. Priorities, which correspond to the order, are assigned to respective branches of the graph.