Abstract: In one embodiment, a method includes receiving, from a client system associated with a user, a first user request that includes a reference to a target object and one or more of an attribute or a relationship of the target object. Visual data including one or more images portraying the target object may then be accessed, and the reference may be resolved to the target object portrayed in the one or more images. Object information of the target object that corresponds to the referenced attribute or relationship of the first user request may be determined based on a visual analysis of the one or more images. Finally, responsive to receiving the first user request, the object information of the target object may be stored in a multimodal dialog state.

Abstract: In some embodiments, a computer-implemented method includes ascertaining a multitier topology representation of an edge cloud network; generating a pseudo node topology representation of the edge cloud network from the multitier topology representation; and utilizing the pseudo node topology representation of the edge cloud network to ascertain minimum-latency pseudo-node-based edge cloud clusters (ECCs), the minimum-latency pseudo-node-based ECCs being utilized to minimize a latency of user requests routed through the edge cloud network from a user of the edge cloud network. In some embodiments of the computer-implemented method, the minimum-latency pseudo-node-based ECCs are ascertained based upon a pseudo-node-based round-trip-times (RTTs) assessment from the user of the edge cloud network, the user requests being routed to the minimum-latency pseudo-node-based ECCs ascertained using the pseudo node topology representation.

Abstract: A ticket exchange server is configured to determine a number of tickets to distribute for an event. The ticket exchange server accesses, for a stadium, training data describing attendance at historical events, historical opponents of a sports team, and a historical win/loss record of the sports team. The ticket exchange server trains a machine-learned model configured to predict an attendance for a future event at the stadium based on an opponent of the sports team at the future event and a current or predicted win/loss record of the sports team. The ticket exchange server selects an event for the sports team against an opponent and determines a predicted attendance using the machine-learned model. The ticket exchange server identifies a number of tickets greater than a capacity of the stadium to make available based on the predicted attendance and distributes the number of tickets to prospective attendees.

Abstract: Systems, apparatuses and methods provide technology that receives different numbers of accesses to a data shard from different regions, where the data shard is a portion of a dataset. The technology identifies, with a machine learning model, access patterns based on the different numbers of accesses, and generates, with the machine learning model, values for the different regions based on the access patterns, where the values represent ranks of the different regions that correspond to a future number of predicted accesses to the data shard from the different regions. The technology determines a subset of the different regions to store the data shard based on the values and stores a replica of the data shard in each of the subset of the regions without scarifying latency.

Abstract: Data characterizing historical skills-based gaming metrics for a first user and historical skills-based gaming metrics for at least one second user is accessed. Using the accessed data and a set of rules, a targeted advertisement to present to the first user is determined. The targeted advertisement specifies at least one skills-based game and a characterization of the at least one second user's historical skills-based gaming metrics. The targeted advertisement is generated. Data characterizing the targeted advertisement is provided. Related apparatus, systems, techniques, and articles are also described.

Abstract: In some embodiments, a metaverse optimization and prioritization enabled cloud-based controller includes a metaverse traffic classification unit; and a metaverse optimization and prioritization unit, wherein based upon the identification and classification of data packets as metaverse data packets, the metaverse optimization and prioritization unit optimizes and prioritizes a metaverse client device and metaverse traffic associated with a metaverse optimization and prioritization enabled network. In some embodiments, the metaverse optimization and prioritization unit optimizes the metaverse client device based upon a quality of experience associated with the metaverse client device. In some embodiments, the metaverse optimization and prioritization unit prioritizes the metaverse traffic based on quality of service management features ascertained utilizing the metaverse optimization and prioritization enabled cloud-based controller.

Abstract: In one embodiment, a method includes receiving a token indicating a data chunk becomes available in a first circular buffer of a pre-determined size from a direct memory access component, determining that a computation is to be performed with data including the data chunk based on the token, and generating one or more addresses corresponding to one or more data chunks within the first circular buffer that are to be retrieved for the computation, where a generated address is subtracted by the pre-determined size of the first circular buffer when the generated address is greater than a pre-determined maximum associated with the first circular buffer, and where the generated address is added by the pre-determined size of the first circular buffer when the generated address is less than a pre-determined minimum associated with the first circular buffer.

Abstract: In particular embodiments, a computing system may access a plurality of images for pre-training a first machine-learning model that includes an encoder and a decoder. Using each image, the system may pre-train the model by dividing the image into a set a patches, selecting a first subset of the patches to be visible and a second subset of the patches to be masked during the pre-training, processing, using the encoder, the first subset of patches to generate corresponding first latent representations, processing, using the decoder, the first latent representations corresponding to the first subset of patches and mask tokens corresponding to the second subset of patches to generate reconstructed patches corresponding to the second subset of patches, the reconstructed patches and the first subset of patches being used to generate a reconstructed image, and updating the model based on comparisons between the image and the reconstructed image.

Abstract: Technology herein provides a method, apparatus and computer readable storage medium for use in video encoding. The technology performs operations including pruning interframe candidate modes, based on one or more criteria, to provide a reduced set of candidate modes for encoding a video block, wherein a candidate mode includes an interframe mode type, a set of reference frame types, and one or more dynamic reference list (DRL) candidates, and wherein pruning interframe candidate modes comprises excluding one or more interframe mode types, determining a rate distortion (RD) cost for each of the candidate modes in the reduced set of candidate modes, selecting a candidate mode from the reduced set of candidate modes, based on the lowest RD cost, as a selected interframe mode, and encoding the video block using the selected interframe mode.

Abstract: Query language statements are generated from natural language statements using a knowledge graph representing one or more databases. The knowledge graph is obtained by creating nodes representing tables and operations referenced by queries to the databases. The data of the databases is evaluated to identify entities and dimensions of entities from among the nodes. The entities are assigned human-understandable labels by an LLM. A natural language statement is converted to a knowledge graph language (KGL) statement and references in the KGL statement are replaced with references to entities in the knowledge graph. The KGL statement is then programmatically converted to a database language statement.

Abstract: Aspects of the present disclosure are directed to an artificial intelligence (“AI”) application running in conjunction with an artificial reality (“XR”) space. The AI Builder responds to user commands, verbal or gestural, to build or edit spaces or objects in space. If the requested object is of a type recognized by the AI Builder, then the AI Builder builds the object from one or more stored templates. The new object's location is determined by the objects that already exist in the user's XR environment and on commands or gestures from the user. If the AI Builder does not recognize the requested object, the user can show an image to the AI Builder, and the AI builds a 3D object in the XR space according to that image. To ease collaboration among users, the AI Builder may present its user interface as a non-player character within the XR world.

Abstract: An apparatus supports single root input/output virtualization (SR-IOV) capable devices. The apparatus includes input/output ports, and SR-IOV capable PCIe devices. Each SR-IOV capable PCIe device has one or more namespaces or controller memory buffers. The SR-IOV capable PCIe device provides one or more physical functions and virtual functions that can access the one or more namespaces or controller memory buffers. A PCIe switch controller communicates with host servers coupled to the input/output ports, and assigns one or more virtual functions to each host device, and enables the host devices to access one or more namespaces or controller memory buffers through the assigned virtual functions.

Abstract: Systems and methods for enabling various devices to remotely interact with cloud-based client applications are provided. A method comprises receiving a first request from a first client device of a user to initiate an interactive session with a cloud-based client application, reserving an application engine for executing the cloud-based client application remotely from the first client device, receiving interaction data from the first client device as the user engages with a first media data associated with the cloud-based client application, modifying the cloud-based client application executing within the application engine that is reserved based on the interaction data received from the first client device, receiving a second request from the first client device to end the interactive session with the cloud-based client application that is modified, and deallocating the application engine that is reserved, wherein the application engine that is reserved is delinked from the first client device.

Abstract: In particular embodiments, a computing system may receive a video comprising a plurality of image frames. The system may generate, for each image frame in the video, an initial depth map using a machine-learning model. The system may compute a misalignment error indicating depth misalignments in initial depth maps using a reprojection technique. The system may generate, for each image frame in the video, an optimized camera pose and a flexible deformation spline associated with the image frame to minimize the misalignment error. The system may generate, for each image frame in the video, a refined depth map by adjusting the initial depth map associated with the frame using the flexible deformation spline associated with the image frame.

Abstract: In one embodiment, a computer-implemented method includes receiving, through a user interface (UI) of an artificial-reality (AR) design tool, a selection of a configurable interface element to place the AR design tool and the UI into a configure phase to configure an AR effect. The computer-implemented method further includes receiving, through the UI of the AR design tool after the AR design tool and the UI are placed into the configure phase in response to selecting the configurable interface element, instructions to add a voice-command module to the AR effect. The computer-implemented method further includes configuring, while the AR design tool and the UI are placed into the configure phase, one or more parameters of the voice-command module. The computer-implemented method further includes generating the AR effect utilizing a particular voice command at runtime based on configured one or more parameters of the voice-command module.

Abstract: In one embodiment, a method for accessing an instruction that is to be executed by a control agent within a computing system may include decompressing the instruction by replacing each of one or more zero-symbol run-length fields in the instruction with as many continuous zero symbols as its corresponding value and removing one or more non-zero-symbol run-length fields from the instruction. The method may also include determining that the instruction is spatial-delta-encoded based on a compression data header associated with the instruction, performing spatial-delta decoding on the instruction in response to the determination by orderly determining a spatial-delta-decoded value of each bit in the instruction, and causing the instruction to be sent to the control agent.

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.

Abstract: A system and method for generating task-specific text by processing multimodal inputs using machine-learning models is provided. The method may include accessing first sets of tokens associated with a desired task and one or more modalities associated with a context of the desired task. The method may further include determining a second set of tokens for each of the one or more modalities using a classifier network associated with the modality. The method may further include generating a number of embedding vectors by mapping the first sets of tokens and the second set of tokens associated with each of the one or more modalities to an embedding space. The method may further include producing a sequence of words addressing the desired task by processing the number of embedding vectors with an encoder-decoder network.