Abstract: In one embodiment, a method includes, by a client system of a first user, presenting a communication interface including a plurality of layers, wherein a first layer of the communication interface includes a video communication of the second user, and wherein a second layer of the communication interface includes a thumbnail view of a video communication of the first user, receiving a request from the second user to initiate a first game, receiving a user input indicative an acceptance of the request to initiate the first game, generating a game container containing the first game in a third layer of the communication interface, reducing the first layer to a thumbnail within the communication interface, and overlaying the first layer onto the third layer, wherein the second layer is closed responsive to the overlaying of the first layer onto the third layer.

Abstract: Apparatuses, methods, and systems for coordinated beamforming in a wireless mesh network, are disclosed. One system includes a network that includes a plurality of nodes connected through wireless links, and a controller. The wireless links including aggressor links and victim links wherein the aggressor links interfere with the victim links. The controller is operative to identify aggressor links and victim links of a group of nodes of the plurality of nodes, coordinate beam scans of the one or more victim receive nodes associated with the victim links of the group, coordinate transmission of one or more aggressor transmit nodes associated with the aggressor links of the group, characterize or receive characterizations of measured interference at the one or more victim receive nodes during the coordinated beam scans, and select beamforming coefficients for the victim receive nodes based at least on the characterizations of the measured interference.

Abstract: A computer-implemented method may include (1) maintaining (a) a filter matrix in a filter cache included in a local memory device (LMD) included in a hardware accelerator, and (b) a plurality of activation matrices corresponding to different rows of an activation volume in an activation cache included in the LMD, (2) for each activation matrix, directing a matrix multiplication unit (MMU) included in the hardware accelerator to execute a matrix multiplication operation (MMU) using the filter matrix and the activation matrix, (3) loading an additional filter matrix into the filter cache, and (4) directing the MMU to execute a plurality of additional MMOs, each additional MMO using one filter matrix included in the filter cache and one activation matrix included in the activation cache, such that the MMU reuses the filter matrix for at least one additional MMO and uses the additional filter matrix for a different additional MMO.

Abstract: Embodiments relate to a sensor system for a brain computer interface (BCI) that enable detection and decoding of brain activity by optical tomography. The sensor system includes an array of pixels arranged as grouped pixel units to provide increased dynamic range. One or more of the grouped pixel units can operate in a saturated mode while providing information useful for decoding brain activity. Furthermore, the grouped pixel units are arranged to enable fast readout by a pixel scanner, thereby increasing detection and decoding ability by systems implementing the sensor design. The grouped pixel units of the sensor system are aligned with optical fibers of an interface to a body region of a user, where the optical fibers can be retained in position relative to the grouped pixel units by an optically transparent substrate that provides mechanical support while minimizing factors associated with divergence of light transmitted through optical fibers.

Abstract: In one embodiment, a method includes receiving a user request from a client system associated with a user, generating a response to the user request which references one or more entities, generating a personalized recommendation based on the user request and the response, wherein the personalized recommendation references one or more of the entities of the response, and sending instructions for presenting the response and the personalized recommendation to the client system.

Abstract: Systems, methods, and non-transitory computer-readable media can determine a first dataset provided by a first party, wherein the first dataset includes a set of vectors that are each associated with a user identifier. A second dataset provided by a second party can be determined, wherein the second dataset includes a set of vectors that are each associated with a user identifier. One or more vectors in the first dataset can be matched to vectors in the second dataset based on a secure multi-party computation without revealing respective graph information of the first party or the second party. Respective mappings between vectors in the first dataset to a set of shared universal identifiers can be provided to the first party. Respective mappings between vectors in the second dataset to the set of shared universal identifiers can be provided to the second party.

Abstract: An electronic device includes one or more microphones that generate audio signals and a wind noise detection subsystem. The electronic device may also include a wind noise reduction subsystem. The wind noise detection subsystem applies multiple wind noise detection techniques to the set of audio signals to generate corresponding indications of whether wind noise is present. The wind noise detection subsystem determines whether wind noise is present based on the indications generated by each detection technique and generates an overall indication of whether wind noise is present. The wind noise reduction subsystem applies one or more wind noise reduction techniques to the audio signal if wind noise is detected. The wind noise detection and reduction techniques may work in multiple domains (e.g., the time, spatial, and frequency domains).

Abstract: Techniques are described that enable a user to edit and customize captions generated by a social networking system, such as transcriptions of an audio clip. In some cases, a social networking system receives, from a first user account, a video and an audio clip associated with the video, and determines that the audio clip contains speech. The social networking system may leverage a speech-to-text component to generate a first text caption based at least in part on the speech in the audio clip. The social networking system provides the first text caption to the first user account, and receives a user input to modify a word included in the first text caption. The social networking system generates a second text caption based at least in part on the user input, and provides the video, including the second text caption, to a second user account.

Abstract: Particular embodiments may receive a request to perform a task to a core network by a user device via an access point. The user device may be authenticated by the core network which comprises one or more network functionality components, and each of the one or more network functionality components may be decomposed into multiple service types. The core network may identify service instances for deployment based on the task. Each of the service instances may belong to one of the multiple decomposed service types. The service instances may be deployed to one or more server machines to accomplish the task by the core network based on resource requirements of the service instances and current resource availability of the one or more server machines.

Abstract: The disclosed computer-implemented method may include (1) provisioning a cloud gaming environment with a plurality of containers that share a single operating system instance, (2) allocating each container within the plurality of containers to a corresponding user, (3) executing, concurrently, within each container within the plurality of containers a corresponding video game instance and (4) streaming, concurrently, from the cloud gaming environment, a video game instance from each container within the plurality of containers to a corresponding client system. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: User action data characterizing action by a player in a game environment executing at a user client is received at a server. The game environment is created by the user client separate from the server. Data characterizing a selected viewing position is received. The selected viewing position is different than a player viewing position. The selected viewing position characterizes a viewing location within the game environment. A recreated game environment is generated from the user action data at the server. A video stream of the recreated game environment is generated. The video stream includes video from a perspective of the selected viewing position. The video stream is transmitted to a viewing client. Related apparatus, systems, articles, and techniques are also described.

Abstract: An online system generates predicted outcomes for a content distribution program that distributes content to users of the online system, the predicted outcome indicating a likelihood for the occurrence of an outcome of a content presentation. The online system transmits the one or more predicted outcomes to the third-party system, and receives prediction improvement data from the third-party system, the prediction improvement data indicating an adjustment to errors in the predicted outcomes based on a prediction by the third-party system. The online system updates the properties of a content distribution program based on the prediction improvement data, the updated content distribution program causing the online system to generate new predicted outcomes based on the prediction improvement data in content presentation opportunities. The online system also transmits content to users of the online system based on the updated content distribution program.

Abstract: An online system receives a content item including an image from an online system user. The online system accesses and applies a trained item detection model to predict a probability that a region of interest within the image corresponds to an item associated with an entity based on a set of pixel values associated with the region of interest. If the probability is at least a threshold probability, the online system accesses and applies a trained quality prediction model to predict a measure of quality of the image based on a set of attributes of the image. If the measure of quality is at least a threshold measure of quality, the online system includes the content item in a page associated with the entity maintained in the online system and/or adds a set of data associated with the item and/or the entity to the content item.

Abstract: An online system receives a content item including an image from a content-providing user and/or receives an interaction with the content item from a viewing user. The online system accesses a machine-learning model that is trained based on a set of images of items associated with an entity and attributes of each image. The online system applies the model to predict a probability that the content item includes an image of an item associated with the entity based on attributes of the image included in the content item. Based on the predicted probability, the online system updates a profile of the user (i.e., the content-providing user and/or the viewing user) to include an affinity for the item. Upon determining an opportunity to present content to the user, the online system selects content for presentation to the user based on the profile and sends the content for presentation to the user.

Abstract: In one embodiment, a method includes receiving a user input comprising a natural-language utterance by an assistant xbot from a client system associated with a user, determining a semantic representation of the user input based on a structural ontology defining a labeling syntax for parsing the natural-language utterance to semantic units comprising actions, objects, and attributes, wherein the semantic representation embeds at least one object within at least one action and declares at least one attribute of the embedded object to be acted upon, sending a request based on the semantic representation to an agent for executing a task corresponding to the user input, receiving results of the executed task mapped to a structure determined by the structural ontology from the agent, and sending from the assistant xbot to the client system instructions for presenting a response based on the results of the executed task.

Abstract: An online system presents content in videos to users. Content providers may value having their content injected into videos from certain sources more than others. This preferences is quantified as a brand value score. The brand value score is determined as a function of user engagement with a source of the video and, to account for brand value, the system performs a two-stage auction. First, the system determines whether to inject any content into a video by determining a distribution of brand value of videos per demand for videos in a previous period and filling a projected demand for the content in a current period to determine a brand value threshold. Then, any videos having a brand value above the threshold are eligible for the second stage of the selection process where the system performs an auction where projected benefit of presenting the content is compared to projected loss.

Abstract: An online system uses a task async engine to coordinate task execution, such as for phased content creation, where the system enables content providers to efficiently create and publish sponsored content in multiple phases. Each phase of the content creation is represented by a draft of the content. Multiple drafts of the content are stored in a repository of the online system for optimizing content delivery and publication, such as quick identification of conflicts in a draft based on analysis of draft fragments of the draft. Objects associated with the sponsored content are stored in an object model. The connections amongst the various draft segments of a draft are stored in a hierarchy structure. Each user associated with a draft has a role, where all of the roles can contribute to the draft. The online system efficiently publishes the drafts and presents the sponsored content to its target users.

Abstract: One or more embodiments of the disclosure provide systems and methods for providing media presentations to users of a media presentation system. A media presentation generally includes a plurality of media segments provided by multiple users of the media presentation system. In one or more embodiments, a user of the media presentation system may share a media presentation with a co-user. The media presentation system can provide a number of features to assist a user in sharing, filtering, and accessing media presentations.

Abstract: A processor system comprises a plurality of processing elements. Each processing element includes a corresponding convolution processor unit configured to perform a portion of a groupwise convolution. The corresponding convolution processor unit determines multiplication results by multiplying each data element of a portion of data elements in a convolution data matrix with a corresponding data element in a corresponding groupwise convolution weight matrix. The portion of data elements in the convolution data matrix that are multiplied belong to different channels and different groups. For each specific channel of the different channels, the corresponding convolution processor unit sums together at least some of the multiplication results belonging to the same specific channel to determine a corresponding channel convolution result data element.

Abstract: Systems, methods, and non-transitory computer-readable media are disclosed for selectively rendering augmented reality content based on predictions regarding a user's ability to visually process the augmented reality content. For instance, the disclosed systems can identify eye tracking information for a user at an initial time. Moreover, the disclosed systems can predict a change in an ability of the user to visually process an augmented reality element at a future time based on the eye tracking information. Additionally, the disclosed systems can selectively render the augmented reality element at the future time based on the predicted change in the ability of the user to visually process the augmented reality element.