Abstract: Traditional messaging applications (such as email, social networking platforms, text, instant messaging, chat, etc.) are ill-equipped for collaboration. For instance, a sender and recipients corresponding regarding a topic are often required to exchange multiple messages. As a result, determining a consensus regarding the topic or summarizing the collaboration is difficult and cumbersome. Moreover, when links are included in messages, recipients are required to navigate away from the messaging application to view the linked content. A messaging application is provided that may be configured to enable users to dynamically collaborate regarding links or data within a sent message or a receive message. In aspects, a collaborative user experience (UX) is provided, which may be an interface within a message that enables real-time collaboration between a sender and one or more recipients regarding message content —without exchanging additional messages or navigating away from the messaging application.

Abstract: A computing device including a processor configured to receive an energy function of a combinatorial optimization problem. The processor may be further configured to compute an inverse temperature lower bound, which may include estimating a maximum change in the energy function between successive timesteps. The processor may be further configured to compute an inverse temperature upper bound, which may include estimating a minimum change in the energy function between successive timesteps. The processor may be further configured to compute the solution to the combinatorial optimization problem at least in part by executing a Markov chain Monte Carlo (MCMC) algorithm over the plurality of timesteps. An inverse temperature of the MCMC algorithm may be set to the inverse temperature lower bound during an initial timestep and may be set to the inverse temperature upper bound during a final timestep. The processor may be further configured to output the solution.

Abstract: This document relates to securing ownership of devices to particular users when the devices are shipped directly from an original equipment manufacturer in order to prevent malicious use of devices that are lost or stolen. A purchaser may purchase a device from an original equipment manufacturer, and as part of the purchasing process, may provide ownership information for the device, which may include a user identifier and an identity provider. The ownership data can be written to firmware, and upon powering on of the device, the device can request a user identifier, which is then validated by the identity provider before allowing operating systems operations to continue on the device.

Abstract: According to examples, an apparatus may include a processor and a memory on which are stored machine-readable instructions that when executed by the processor, may cause the processor to obtain an encryption key from a user. The processor may identify session activity data during a proxy session of the user and may encrypt the identified session activity data using the encryption key obtained from the user. The processor may store the encrypted session activity data.

Abstract: Examples relate to managing power consumption of a stylus haptic feedback component prior to actuation. In one example, power is transmitted to a haptic circuit and a first haptic predicter value corresponding to a first user interaction with the stylus is determined. A weighted first haptic predicter value is generated by weighting the first haptic predicter value. A second haptic predicter value corresponding to a second user interaction is determined, and a weighted second haptic predicter value is generated by weighting the second haptic predicter value. At least the weighted first and second haptic predicter values are combined to generate a combined weighted predictive result, which is compared to a haptic predictive threshold value. On condition that such comparison yields a haptic predictive result, power continues transmitting to the haptic circuit. On condition that such comparison yields a non-haptic-predictive result, power ceases transmitting to the haptic circuit.

Abstract: The description relates to hinged devices, such as hinged computing devices. One example can include a first portion having a first display and a second portion having a second display and a hinge assembly rotatably securing the first and second portions. The hinge assembly can define rotation of the first portion by a first pair of hinge guide pins riding in a first guide track and rotation of the second portion by a second pair of hinge guide pins riding in a second guide track.

Abstract: A quantum computing device including a doubly controlled iX (CCiX) circuit. The CCiX circuit may be configured to, in a preparation stage, prepare a plurality of magic states. The CCiX circuit may be further configured to receive a plurality of input qubit states including a first control qubit state, a second control qubit state, and a target qubit state. In an execution stage, the CCiX circuit may be further configured to perform a CCiX operation on the target qubit state at least in part by performing a plurality of local joint measurements. At least a subset of the plurality of local joint measurements may be performed between the plurality of magic states and a plurality of auxiliary qubits. Performing the CCiX operation may further include performing a plurality of remote joint measurements of the input qubit states and a plurality of interface qubits included among the plurality of auxiliary qubits.

Abstract: A data processing system implements a method for training machine learning modes, including receiving a set of one or more unlabeled documents associated one or more first categories of documents to be used to train machine learning models to analyze the one or more unlabeled documents, and fine-tuning a first machine learning model and a second machine learning model based on the one or more unlabeled document to enable the first machine learning model to determine a semantic representation of the one or more first categories of document, and to enable the second machine learning model to classify the semantic representations according to the one or more first categories of documents, the first machine learning model and the second machine learning model having been trained using first unlabeled training data including a second plurality of categories of documents that do not include the one or more first categories of documents.

Abstract: A data processing system is implemented for receiving speech data for a plurality of languages, and determining letters from the speech data. The data processing system also implements normalizing the speech data by applying linguistic based rules for Latin-based languages on the determined letters, building a computer model using the normalized speech data, fine-tuning the computer model using additional speech data, and recognizing words in a target language using the fine-tuned computer model.

Abstract: A data processing system implements a dynamic knowledge distillation process including dividing training data into a plurality of batches of samples and distilling a student model from a teacher model using an iterative knowledge distillation. The process includes instantiating an instance of the teacher model and the student model in a memory of the data processing system and obtaining a respective batch of training data from the plurality of batches of samples in the memory. The process includes training the teacher and student models using each of the samples in the respective batch of the training data, evaluating the performance of the student model compared with the performance of the teacher model, and providing feedback to student model to adjust the behavior of the student model based on the performance of the student model.

Abstract: A computing system measures an amount of carbon that is emitted as a consequence of the execution of individual application processes of an operation. The computing system can then leverage these measurements to modify how the application processes are run. In operation, the computing system deploys an application process in a container of a particular worker node of a cluster in a container orchestration platform. The technology further deploys an agent (e.g., a daemon) on all of the worker nodes of the cluster, including the particular worker node. The agent includes an energy-managing mechanism that is given privileges to measure the amount of energy consumed by the application process within the particular node. The computing system can use this energy measure, together with applicable carbon intensity information, to compute the amount of carbon emitted by the application process.

Abstract: A quantum computing device is provided, including a table lookup circuit configured to receive a first table lookup input and a second table lookup input. The table lookup circuit may be further configured to perform a first table lookup operation on the first table lookup input and a second table lookup operation on the second table lookup input in parallel such that a combined table lookup output is written to a combined output register. The combined table lookup output may include a plurality of first table lookup output qubits of the first table lookup operation and a plurality of second table lookup output qubits of the second table lookup operation.

Abstract: A data processing system implements obtaining attention matrices from a first machine learning model that is pretrained and includes a plurality of self-attention layers. The data processing system further implements analyzing the attention matrices to generate a computation graph based on the attention matrices. The computation graph provides a representation of behavior of the first machine learning model across the plurality of self-attention layers. The data processing system is further implements analyzing the computation graph using a second machine learning model. The second machine learning model is trained to receive the computation graph to output model behavior information. The model behavior information identifying which layers of model performed specific tasks associated with generating predictions by the first machine learning model.

Abstract: Systems and methods are provided for alleviating congestion of data traffic in a wireless network in a cell site. The methods determine congestion levels for respective data connections from mobile computing devices to a network through a cell tower in a cell site. The method further determines a congestion level associated with the cell site based on a weighted aggregate of varying congestion levels associated with the respective data connections in the cell site. The disclosed technology includes flow policies associated with congestion levels. The flow policies determine weights used for determining the weighted aggregate and maximum network bandwidth enforced for the respective data connections. The disclosed technology identifies data connections that consume abnormally high amount of network bandwidth and throttles with bias against data traffic associated with these data connections with heavy traffic.

Abstract: A knowledge injection model for generative commonsense reasoning. In examples, an encoder-decoder model is used to generate a model output (204) a plausible description for a set of concepts. A prototype (218) is generated from an in-domain or out-of-domain knowledge corpus, which is further used as input (202) for the encoder-decoder model. Concept input tokens and prototype input tokens are scaled to limit potential skew that may be introduced by the prototype (218). Additionally, position indicators are generated for each input token, which indicate the relative position each respective input token as compared to other input tokens. As such, when decoding the scaled encoded input tokens, the decoder (214) may be more attuned to the scenario bias that is introduced by the prototype (218) when generating a model output (204). Thus, the encoder-decoder model need not rely solely on the set of concepts when generating the model output (204).

Abstract: A system and method for automatically generating help documentation for an application includes examining telemetry data associated with a plurality of users' use of the application and identifying from the telemetry data an intended action for which a help documentation should be generated. Upon identifying the intended action, an action path in the application for arriving at the intended action may be determined where the action path includes one or more actions taken before arriving at the intended action and an action label for each of the actions may be identified. The action path and the action label may be provided to a trained machine-learning (ML) model for automatically generating the help documentation and in response, an automatically generated help documentation may be received as an output.

Abstract: Aspects of the present disclosure relate to a providing related content recommendations in response to a user search query by supervising the training of pair embeddings using contrastive learning and pairwise co-click signals. The approach combines a two tower model architecture with a cascaded multilayer perceptron model to enable the adoption of variable combinations of input features and more representative learned pair embeddings. The learned embeddings undergo supervised contrastive loss training to generate a related content recommendation model, which is subsequently evaluated using both online and offline metrics. The related content recommendation model can provide results to search queries that improve recommendation quality and increase user engagement, thereby ultimately enhancing long term user experience.

Abstract: A computer device includes a processor configured to simulate a virtual environment based on a set of virtual environment parameters, and perform ray tracing to render a view of the simulated virtual environment. The ray tracing includes generating a plurality of rays for one or more pixels of the rendered view of the simulated virtual environment. The processor is further configured to determine sub-pixel data for each of the plurality of rays based on intersections between the plurality of rays and the simulated virtual environment, and store the determined sub-pixel data for each of the plurality of rays in an image file.

Abstract: An electronic device comprises a touch sensor, a movement sensor, and recalibration logic. The touch sensor provides input to the electronic device. The movement sensor has an output that varies in dependence on movement of the electronic device, and the recalibration logic is configured to vary a rate of successful recalibration of the touch sensor based at least partly on the output of the movement sensor. The successful recalibration comprises replacement of a stored calibration mapping of each of a plurality of (X, Y) coordinates of the touch sensor to a corresponding touch-free capacitance.

Abstract: Aspects of the present disclosure relate to sustainability-aware data center workload management. In examples, a data center with which to process a workload is determined from a set of candidate data centers. The set of candidate data centers may include data centers that are geographically dispersed to account for variability in the availability of renewable energy, among additional or alternative environmental considerations. For example, the set of candidate data centers may be generated such that at least one candidate data center is in a location that experiences sunlight at any given time. As a result, a preferred data center from the set of data centers may vary based on environmental conditions (e.g., a time of day, a current season, and/or a weather forecast), such that workload processing is managed using the set of candidate data centers to reduce the associated environmental impact accordingly.

Abstract: According to implementations of the subject matter described herein, there is proposed a solution for automatic analysis of a difference between multi-dimensional datasets. In this solution, an analysis request is received for a first dataset and a second dataset, each of which including data items corresponding to a plurality of dimensions. In response to the analysis request, data items corresponding to a first dimension in the first and second datasets are compared. Based on the comparison, a first set of influence factors associated with the first dimension are determined, each influence factor indicating a reason for a difference between the first and second datasets from a respective perspective. An analysis result related to the difference between the first and second datasets is presented based on the first set of influence factors. In this way, it is possible to achieve automatic and efficient analysis of the difference between the different datasets.

Abstract: A system and method for executing SQL statements includes receiving an SQL statement for comparing two trendsets over a relation using a scoring function, each of the trendsets including one or more trends, each of the trends being designated by a constraint and a grouping-measure combination, wherein comparing the trendsets includes identifying trend pairs for comparison, each of the trend pairs including a trend from the each of the trendsets having a common grouping-measure combination. The SQL statement is transformed into a basic plan of existing logical operators for performing the SQL statement. A set of sub-plans is determined based on the basic plan. Pairs of sub-plans are merged to generate a set of merged sub-plans. A cost for each of the merged sub-plans is determined. The merged sub-plan having the lowest cost is used to execute the SQL statement.

Abstract: Systems and methods for synchronizing a source document repository with a target document repository include monitoring documents of the source document repository to detect changes to the documents. A last processed time is stored in association each changed document that indicates a time when the document was last processed. A transaction item is added to a synchronization queue each time a document is changed, and a queue time is stored in association with each of the transaction items that indicates when the transaction item was added to the queue. The transaction items are processed by comparing the queue time of the transaction item with the last processed time of the document pertaining to the transaction item. If the queue time is earlier than the last processed time, the transaction item is ignored. If the queue time is not earlier than the last processed time, the document pertaining to the transaction item is synchronized to the target document repository.

Abstract: Aspects of the present disclosure relate to providing, based on a partial query string, a plurality of autosuggestions that are diverse in nature such that the user is more likely to see the preferred complete query terms and therefore more likely to select one of the preferred suggestions which will increase search efficiency. As described herein, such functionality relates to generating cluster groups of candidate suggestions, each cluster including sub-topics, then performing the search based on a selected cluster or sub-topic. To generate the cluster groups, systems and methods, as described herein, analyze the similarity between candidate suggestions as well as the popularity of generated sub-topics. The cluster groups and sub-topics may be displayed visually, and in certain embodiments the cluster groups are ordered vertically from top to bottom and aligned to the left side of the display, while sub-topics are ordered horizontally from left to right following the cluster label.

Abstract: A foldable computing device comprises a first portion, a second portion, a hinge assembly rotatably coupling the first portion to the second portion, and a flexible display. The flexible display extends at least partially across the first portion and at least partially across the second portion, and spans the hinge assembly between the first portion and the second portion. The hinge assembly includes a hinge core, a first linkage component rotatably coupling the hinge core to the first portion and forming a portion of a backing support for the flexible display, and a first hinge guide rotatably coupling the hinge core to the first portion. The hinge assembly further includes a second linkage component rotatably coupling the hinge core to the second portion and forming another portion of the backing support for the flexible display, and a second hinge guide rotatably coupling the hinge core to the second portion.

Abstract: A foldable computing device comprises a first portion, a second portion, a hinge assembly rotatably coupling the first portion to the second portion, and a flexible display. The hinge assembly includes a hinge core, a timing element, a first linkage component rotatably coupling the hinge core to the first portion, a first hinge guide rotatably coupling the hinge core to the first portion, a second linkage component rotatably coupling the hinge core to the second portion, and a second hinge guide rotatably coupling the hinge core to the second portion. The first linkage component defines one or more helical surfaces that interface with the timing element, and the second linkage component defines one or more helical surfaces that interface with the timing element. The timing element coordinates rotation of the first linkage component with rotation of the second linkage component.

Abstract: This document relates to a process for supporting the management of a variety of types of deployed devices. Administrators utilizing enterprise services can provide generic configuration data using configuration templates, which can be provided to a management server. The management server can then precompute device-specific configuration settings and resolve any conflicts that may arise based on the configuration templates. The configuration templates can also include placeholders for secret values, and once a managed device checks in to the management server, the secret values can be retrieved from an applicable enterprise service and provided to the managed device at the time of applying the configuration template.

Abstract: A server system is provided that includes one or more compute nodes that include at least one processor and a host memory device. The server system further includes a plurality of solid-state drive (SSD) devices, a local non-volatile memory express virtualization (LNV) device, and a non-transparent (NT) switch for a peripheral component interconnect express (PCIe) bus that interconnects the plurality of SSD devices and the LNV device to the at least one processor of each compute node. The LNV device is configured to virtualize hardware resources of the plurality of SSD devices. The plurality of SSD devices are configured to directly access data buffers of the host memory device. The NT switch is configured to hide the plurality of SSD devices such that the plurality of SSD devices are not visible to the at least one processor of each compute node.

Abstract: In examples, a system for using interchangeable non-compute resources is provided. The system includes at least one processor and memory storing instructions that, when executed by the at least one processor, cause the system to: receive a first key corresponding to a first non-compute resource, define a first connection with the first non-compute resource based on the first key, receive a second key corresponding to a second non-compute resource, define a second connection with the second non-compute resource based on the second key, receive an indication corresponding to a selection of the first non-compute resource, and configure the system to interface with the first non-compute resource at a deployment of a generated application.

Abstract: In various embodiments, a method for processing video streams is described. A plurality of video streams for transmission to a display device are received. The plurality of video streams have respective initial image quality levels. An estimated gaze location of a user of the display device is estimated. At least one video stream of the plurality of video streams is processed to have a modified image quality level based on the estimated gaze location. The modified image quality level is less than a corresponding initial image quality level. The plurality of video streams are transmitted to the display device.

Abstract: Aspects of the present disclosure relate to a personalized agent service that generates and evolves customized agents that can be instantiated in-game to play with users. Machine learning models are trained to control the agent's interactions with the game environment and the user during gameplay. A user may request that a personalized agent join the user's gameplay session. The user device sends a request for the personalized agent to a game platform. The game platform determines whether the user has a license to execute a second instance of the game. When the user has a license to execute a second instance of the game, the second instance of the game may be executed on the user device. Information received from a personalized agent service is used to instantiate a personalized agent in the second instance of the game.

Abstract: A method of training an index filter for an index tuning system includes receiving a plurality of different workloads and a plurality of different databases, each database including different tables and each workload including a plurality of queries; generating labeled training by making optimizer calls to a query optimizer using query and index configuration pairs from the plurality of databases and the plurality of workloads; training an index filter model to identify signals in the labeled training data, the signals being indicative of a potential performance improvement associated with using an index configuration for a given query; training the index filter model to learn rules over the signals for identifying spurious indexes; and storing the index filter model in a memory.

Abstract: A computing system uses AES-XTS encryption to encrypt data of a first part of first data stream using a tweak key, a data key, an initial tweak value, in a first encryption session, store the encrypted first part, then encrypts a second part of the first data stream in a second encryption session commenced after the termination of the first encryption session; and store the encrypted second part in the encrypted data store. The second part of the first data stream is encrypted using a modified tweak value computed based on the initial tweak value, the tweak key, and a block index of a last cipher block of the first part of the first data stream.

Abstract: Examples are disclosed herein that relate to determining whether an imaged subject is real or spoofed. One example provides a computing system, comprising, a camera, a light pattern source configured to output a light pattern, a logic subsystem, a storage subsystem storing instructions executable by the logic subsystem to capture, via the camera, an image of a subject illuminated by the light pattern emitted by the light pattern source, determine, based at least upon a contrast of the light pattern in the image, whether the subject is real or a spoof, based at least upon determining that the subject is real, perform an action on the computing system, and based at least up on determining that the subject is a spoof, not perform the action on the computing system.

Abstract: A computer-implemented method is disclosed. A search query of a text transcription is received. The search query includes a word or words having a specified spelling. A sequence of search phonemes corresponding to the specified spelling is generated. A sequence of transcript phonemes corresponding to the text transcription is generated from the text transcription. A search alignment in which the sequence of search phonemes is aligned to a transcript phoneme fragment is generated. Based at least on the search alignment having a quality score exceeding a quality score threshold, the transcript phoneme fragment and an associated portion of the text transcription is determined to result from an utterance of the specified spelling in an audio session corresponding to the text transcription. A search result indicating that the transcript phoneme fragment and the associated portion of the text transcription is determined to have resulted from the utterance is output.

Abstract: Examples of the present disclosure describe improved systems and methods for disaster recovery for edge devices. In one example implementation, a current device configuration for a first device is received. The current device configuration comprises a device state and a workload configuration of the first device. A bootstrapping package for the first device is generated based on the current device configuration for the first device. Generating the bootstrapping package comprises segmenting the device state and the workload configuration into a namespace. The bootstrapping package is provided to a second device. The bootstrapping package is configured to be automatically installed on the second device.

Abstract: Aspects of the present disclosure relate to an audio communication device with novel visual indications and adjustable muting. In examples, an audio communication device is provided. The audio communication device includes at least one processor, and memory storing instructions that, when executed by the at least one processor, cause the audio communication device to perform a set of operations. The set of operations include generating a first visual indicator. The first visual indicator corresponds to audio of one or more users. The set of operations further include receiving a user-input, muting a microphone region of the audio communication device, based on the user-input, and generating a second visual indicator. The second visual indicator replaces at least a portion of the first visual indicator. The second visual indicator corresponds to the muted region. The first visual indicator is indicative of a quality of the audio.

Abstract: Systems and methods are provided for generating a keyword sequence from an input query. A first text sequence corresponding to an input query may be received and encoded into a source sequence representation using an encoder of a machine learning model. A keyword sentence may then be generated from the source sequence representation using a decoder of the machine learning model. The decoder may generate a modified generation score for a plurality of prediction tokens, wherein the modified generation score is based on the respective prediction token generation score and a maximum generation score for a suffix of each prediction token. The decoder may then select the prediction token of the plurality of prediction tokens based on the modified generation score, and add the selected prediction token to the previously decoded partial hypothesis provided by the decoder.

Abstract: Examples of the present disclosure describe systems and methods for notifying users of content creation related to existing content expertise. In examples, content associated with a data domain is evaluated to determine the topics associated with the content. User activity for the data domain is monitored to determine user knowledge levels on various topics. User collaboration circles are determined based on interactions between users. When new content is detected, topics in the new content are evaluated and users having knowledge on the topics are identified based on the determined knowledge levels. The users having knowledge on the topics are ranked based in part on their collaboration circles. At least a portion of the ranked users having knowledge on the topics are notified of the new content creation and/or the users creating the new content are notified of the ranked users having knowledge on the topics.

Abstract: Examples of the present disclosure describe systems and methods for maintaining stale files to minimize computing costs. In examples, a determination is made regarding whether a hydrated file is eligible to remain temporarily stale. If the hydrated file is determined to be eligible to remain temporarily stale, the hydrated file is caused to remain stale at a time when a content refresh for the hydrated file is pending. While the hydrated file remains stale, a set of rules for determining whether the hydrated file should be refreshed is evaluated. If, based on the set of rules, it is determined that the hydrated file should be refreshed, the hydrated file is refreshed.

Abstract: Systems and methods for creating and deploying jobs in a cloud-based service include receiving a job definition defining job attributes pertaining to a job to be deployed in a cloud-based service. The job definition is processed to determine whether the job definition complies with predefined rules for job creation in the cloud-based service. Based on the determination of whether the job definition complies with the predefined rules for job creation in the cloud-based server, the job may be validated or not validated in the cloud-based service. Deployment policies are determined for validated jobs based on the job definition for the validated job.

Abstract: A method for sparse matrix multiplication comprises receiving a first block having M elements in a first dimension, and parsing the first block of M elements into a first set of B sub-blocks including MB elements in the first dimension. A first sparsity mask having S % sparsity is applied to the first block of elements, such that each of the first set of B sub-blocks has S % sparsity. A second block is received having M elements in a second dimension, and is parsed into a second set of B sub-blocks that include MB elements in the second dimension. A second sparsity mask having S?% sparsity is applied to the second block of elements, such that S?% of the second set of B sub-blocks have 100% sparsity and (100?S?)% of the second set of B sub-blocks have 0% sparsity. The first and second blocks are then matrix multiplied.

Abstract: The present disclosure relates to a system for optimizing a trial of software product within a budget. In particular, the disclosed technology estimates a resource value of product acquisition for a product and determines a trial budget for the product. A software provider offers the first trial of the product to the user for a first trial period based on the trial budget and a resource cost associated with providing the product and collects data associated with the first trial of the product. In response to the first trial not resulting in the product acquisition, a subsequent trial optimizer uses a model for evaluating the collected data and determines a parameter for optimizing a second trial of the product for the user within a remaining trial budget. Based on the at least one parameter, the system offers the second trial of the product to the user.

Abstract: A method for scheduling one or more of a plurality of calendar events comprises generating a plurality of partial candidate calendars for an attendee. Each partial candidate calendar includes, for each of one or more calendar events, a timeslot selected from one or more timeslot options for that calendar event. An opportunity cost is calculated for each partial candidate calendar. At least some of the partial candidate calendars are queued for further development in which each queued partial candidate calendar is a parent node in a calendar graph. The further development includes adding a calendar event to the parent node to thereby generate another partial candidate calendar or a completed candidate calendar. The queuing is sequenced based on the opportunity cost for each partial candidate calendar. At least a portion of the plurality of calendar events are scheduled using one or more completed candidate calendars generated from the queuing.

Abstract: Systems and methods for implementing changes made to source code of a program at runtime are provided. In particular, a computing device may execute the source code in a runtime environment and detect changes made to the source code at runtime in the runtime environment. In response to detecting the changes, the computing device may determine if the changes were made to a type in the source code identified as reloadable. In response to determining that the changes were made to the type in the source code identified as reloadable, the computing device may create a new version of the type and perform implementation of the changes to the source code at runtime using the new version of the type.

Abstract: Aspects of the present disclosure relate to allocating RAN resources among RAN slices using a machine learning model. In examples, the machine learning model may determine an optimal RAN resource configuration based on compute power needs. As a result, RAN resource allocation generation and compute power requirements may improve, even in instances with changing or unknown network conditions. In examples, a prediction engine may receive communication parameters and/or requirements associated with service-level agreements (SLAs) for applications executing at least partially at a device in communication with the RAN. The RAN may generate one or more RAN resource configuration for implementation among RAN slices. Upon a change in network conditions or SLA requirements, an optimal RAN configuration may be determined in terms of required compute power.

Abstract: Examples are disclosed that relate to performing computer-based well being assessments and proactively performing computer-based mitigation operations to improve a user's well being. In one example, a computing system comprises a network communication subsystem, an attribution machine, a well being assessment machine, and a mitigation machine. The network communication subsystem is configured to communicate with a plurality of user computers. The attribution machine is configured to attribute, to a user account, computing information that the network communication subsystem receives from a user computer. The well being assessment machine is configured to progressively update a well being score over time for the user account based at least on the computing information. The mitigation machine is configured to perform a mitigation operation associated with the user account based at least on an above threshold rate of decrease of the well being score.

Abstract: Aspects of the present disclosure relate to a personalized agent service that generates and evolves customized agents that can be instantiated in-game to play with users. Machine learning models are trained to control the agent's interactions with the game environment and the user during gameplay. As the user continues to play with the agent, the one or more machine learning models develop gameplay styles for the agent that complement the user's preferred playstyle, incorporate the user's preferred strategies, and is generally customized for interaction with the user. The agent personalization data generated during gameplay is stored by the service. An application programming interface is provided by the personalized agent service. Using the API, games can import agent personalization data in order to customize in-game non-player characters (NPCs), thereby customizing the in-game NPCs in accordance with the user's preferences.

Abstract: A method for utilizing computing resources in a vRAN is described. A predicted resource load is determined for data traffic processing of wireless communication channels served by the vRAN using a trained neural network model. The data traffic processing comprises at least one of PHY data processing or MAC processing for a 5G RAN. Computing resources are allocated for the data traffic processing based on the predicted resource load. Wireless parameter limits are determined for the wireless communication channels that constrain utilization of the allocated computing resources using the trained neural network model, including setting one or more of a maximum number of radio resource units per timeslot or a maximum MCS index for the wireless parameter limits. The data traffic processing is performed using the wireless parameter limits to reduce load spikes that cause a violation of real-time deadlines for the data traffic processing.

Abstract: A network system for a data center is described in which an access node sprays a data flow of packets over a logical tunnel to another access node. In one example, a method comprises establishing, by a plurality of access nodes, a logical tunnel over a plurality of data paths across a switch fabric between a source access node and a destination access node included within the plurality of access nodes, wherein the source access node is coupled to a source network device; and spraying, by the source access node, a data flow of packets over the logical tunnel to the destination access node, wherein the source access node receives the data flow of packets from the source network device, and wherein spraying the data flow of packets includes directing each of the packets within the data flow to a least loaded data path.