Abstract: A technique is described for compressing input information fed to a machine-trained generative model. The technique includes: receiving original input information having a plurality of sentences; performing word-level encoding of the original input information using a first part of a machine-trained transformer model, to provide word-level encoded information; performing sentence-level encoding of the word-level encoded information using a second part of the machine-trained transformer model, to provide scores associated with the first plurality of sentences; selecting a subset of the sentences in the original input information based on the scores, to provide modified input information; and providing the modified input information to the machine-trained generative model. The operation of word-level encoding performs parallel processing on portions of the original input information.

Abstract: In some examples of the disclosed technology, a method comprises etching titanium dioxide selectively in the presence of aluminium oxide using an etchant. The etchant can be an aqueous solution of a base and a hydroperoxide. In some examples, the etchant is effective for removing titanium dioxide without destroying an aluminium oxide layer. Examples of an etchant for etching titanium dioxide are disclosed, where the etchant is an aqueous solution of ammonia and hydrogen peroxide, and in some examples the molar ratio of the ammonia to the hydrogen peroxide is in the range 1:34 to 1:38. Methods of etching titanium dioxide disclosed herein can be used to fabricate semiconductor-superconductor hybrid devices.

Abstract: A computing system is provided for selective memory retrieval. The computing system includes processing circuitry configured to provide access to a plurality of memory banks, cause an interaction interface for a trained generative model to be presented, receive, via the interaction interface, an instruction from the user for the trained generative model to generate an output, extract a context of the instruction, generate a memory request including the context and the instruction, input the memory request into a plurality of memory retrieval agents respectively coupled to the plurality of memory banks to retrieve a plurality of relevant memories, generate a prompt based on the retrieved relevant memories and the instruction from the user, provide the prompt to the trained generative model, receive, in response to the prompt, a response from the trained generative model, and output the response to the user.

Abstract: Data may contain personal information and be subject to privacy requirements. The data may be encrypted and only a secure enclave may be able to decrypt the encrypted data. The secure enclave may be used to generate a report based on the encrypted data and a first set of added noise. The report may be subject to audit requirements and satisfy a differential privacy guarantee. The encrypted data may be stored for a first period. After the first period, the secure enclave may be used to generate a private synopsis based on the encrypted data and a second set of added noise. The private synopsis may satisfy the differential privacy guarantee. The private synopsis may be encrypted and only the secure enclave may be able to decrypt the encrypted private synopsis. The encrypted data may be expunged, and the encrypted private synopsis may be retained for a second period.

Abstract: A computing system is provided, comprising at least one processor configured to receive a user interaction history of a user, extract memories from the user interaction history, consolidate the memories into memory clusters, cause a prompt interface for a trained model to be presented, receive, via the prompt interface, an instruction from the user for the trained model to generate an output, generate a prompt based on the memory clusters and the instruction from the user, provide the prompt to the trained model, generate, in response to the prompt, a response via the trained model, and output the response to the user.

Abstract: According to one aspect, a computing system is provided that includes processing circuitry configured to receive input data from multiple interaction modalities of a user, generate a multi-interaction-modality user interaction history from the input data, and extract memories from the multi-interaction-modality user interaction history using a trained memory-extracting generative model. The memories include natural language text descriptions of interactions in the user interaction history generated by the trained memory-extracting generative model. The processing circuitry is further configured to store the memories in file storage having an associated database with a vector search interface configured to receive memory retrieval queries.

Abstract: Computing devices and methods of upscaling video data are disclosed. In one example a method of upscaling video data comprises generating first resolution video data comprising a plurality of tiles that each comprise a plurality of pixels. The method determines whether a first tile of the plurality of tiles matches a previously-received version of the first tile. At least on condition that the first tile does not match the previously-received version of the first tile, the first tile is upscaled from a first resolution to a second resolution greater than the first resolution. The method determines whether a second tile of the plurality of tiles matches a previously-received version of the second tile. At least on condition that the second tile matches the previously-received version of the second tile, the method refrains from upscaling the second tile from the first resolution to the second resolution.

Abstract: A method for fabricating an image sensor comprises: forming an array of sensor elements on a sensor-wafer substrate; forming a readout circuit on the sensor-wafer substrate; forming a plurality of signal lines between the array of sensor elements and the readout circuit; forming a solid-state cooler between the array of sensor elements and the readout circuit; bonding a carrier-wafer substrate to an epitaxial structure of the sensor-wafer substrate; etching the carrier-wafer substrate in the thermal-barrier zone to form a carrier-wafer trench between the array of sensor elements and the readout circuit; reducing the thickness of the sensor-wafer substrate; and etching the sensor-wafer substrate in the thermal-barrier zone to form a sensor-wafer trench between the array of sensor elements and the readout circuit.

Abstract: Methods, systems, apparatuses, and computer-readable storage mediums described herein enable executable code of a hardware security platform (HSP) circuit to communicate with a hypervisor in a separate processor. The hypervisor generates and manages virtual machines. The HSP code comprises trusted platform module (TPM) logic, that processes TPM commands received via the hypervisor, and in response to the processing, communicates security information (e.g., measurements, keys, authorization data) with the virtual machines via the hypervisor. The TPM logic receives security information related to a virtual machine from the hypervisor and stores the security information in non-volatile memory of the HSP circuit, where security information from a particular VM is distinguishable from security information from another VM in the HSP memory.

Abstract: According to examples, an apparatus may include a processor and a memory on which is stored machine-readable instructions that when executed by the processor, may cause the processor to determine that an entity was granted an anomalous role assignment to a managed environment. The processor may also, based on the determination that the role assignment of the entity is anomalous, identify at least one indicator associated with the role assignment, determine an indicator value corresponding to the identified at least one indicator, and determine whether the indicator value exceeds a predefined threshold value. The processor may, based on a determination that the indicator value exceeds the predefined threshold value, output an alert indicating that the role assignment is suspicious.

Abstract: A computing device is provided. The computing device may include a first camera configured to capture a primary image sequence of a scene, and a second camera configured to substantially concurrently capture a secondary image sequence of the scene. The computing device may include a processor configured to execute instructions stored in memory to obtain a depth map for an image of the primary image sequence, and based on the depth map, generate a reference frame of background pixels from the primary image sequence. The processor may be configured to execute the instructions to detect an intrusive object in the secondary image sequence, replace the intrusive object in the primary image sequence with corresponding background pixels of the reference frame to generate a modified primary image sequence, and output on a display at least one image of the modified primary image sequence.

Abstract: A computing device comprises a cover, a chassis, a cover closure sensor, and a cooling fan. An airflow detector and a thermal sensor are located within the chassis. The computing device comprises a processor and a memory storing instructions executable by the processor to detect a thermal trip condition using at least signals from the airflow detector, the thermal sensor, and the cover closure sensor. A cooling action is performed at least on condition of (1) determining that a velocity of an airflow within at least a portion of the chassis is less than or equal to a threshold velocity, (2) determining that the cover is closed, and (3) determining that a temperature of the computing device is greater than or equal to a threshold temperature.

Abstract: In an example embodiment, machine learning is used to train a machine-learned model that projects each entity, title pair into a single number, called a seniority score, to represent the career progression needed for that position. For example, company A's “software engineer” and company B's “senior software engineer” can be represented as two separate numbers, one being p (company A, software engineer) and the other being p (company B, senior software engineer) on the same axis. This allows a comparison to be made about the absolute levels of each title despite their potential different meanings at different entities.

Abstract: Systems and methods are provided for implementing a universal search indexer for enterprise and cloud accessible websites. A universal search indexer, using a crawling agent, crawls a target website and/or web documents in the target website, which includes a plurality of webpages including at least one of one or more static webpages or one or more dynamic webpages. The universal search indexer extracts website content and/or web documents as the target website is being crawled, and ingests the extracted website content and/or web documents within a data store, by indexing the extracted website content and/or web documents in a search index of the data store. The extracted website content and/or web documents are indexed to be searchable and refinable using a search engine, the extracted website content and/or web documents being retrievable via the search engine.

Abstract: Systems and methods relating to full motion video (FMV) routing in one-way transfer (OWT) systems are described herein. The present technology modifies or adds packetized elementary streams (PESs) of video streams, that are sent from the low-trust side of the OWT system, with a global unique identifier (GUID) that is used as an identifier to determine a particular destination on the high-trust side of the OWT system. The enriched video stream is then transmitted through an OWT system that provide high reliability for the enriched video stream. When the enriched video stream is received on the high-trust side, the GUID from the PES is extracted and used to identify destination addresses for destination devices in the high-trust computing environment. The video stream is then delivered to the destination devices having the corresponding destination addresses.

Abstract: Systems and methods are provided for implementing a cluster-wide root secret (“CWRS”) key for distributed node clusters. In a multi-node cluster, a leader node has a leader node security system that generates the CWRS key, which is a common secret key for all workloads (e.g., containers or VMs) in the multi-node cluster. The leader node security system encrypts the generated CWRS key using a public key and/or a bootstrap key received from a non-leader node that requests the CWRS key. In examples, the leader node security system signs the encrypted CWRS key using its private key for subsequent verification, by the requesting non-leader node, that the CWRS key was generated by the leader node security system. The CWRS thus encrypted can be securely sent to the requesting non-leader node for subsequent encryption or decryption of secret data by the security system of the non-leader node.

Abstract: In one example of the disclosed technology, a method of fabricating a device comprises forming a patterned layer of a material on a surface of a substrate by depositing the material through a stencil mask, and forming a passivating layer over the patterned layer and the substrate surface in a sealed apparatus, the substrate being maintained under a vacuum until after the passivating layer has been formed. In some examples, the passivation is performed by oxidising a deposited aluminium layer within a deposition chamber. In some examples, the method can be used for fabricating hybrid semiconductor-superconductor devices, such as Majorana zero mode (MZM) nanowire structures for topological quantum bits.

Abstract: Aspects of the present disclosure relate to adverse user behavior detection and mitigation. In examples, a user account of a social platform may be trusted or untrusted. If the user account is untrusted, activity of the user account may be buffered so as to not directly affect social metrics of the social platform. For example, if the untrusted user account follows a target user account, the untrusted user account may be added to a separate set of followers or otherwise separated from a set of trusted followers of the target user account. Eventually, each user account in the separate set is evaluated to determine whether the user account has transitioned to a trusted user. If the user account is now trusted, it may be migrated to the set of trusted followers. However, if the user account is untrusted, it may be removed from the separate set, thereby reverting the activity.

Abstract: Embodiments of the disclosed technologies include generating a search prompt based on an input portion of an online dialog involving a user of a computing device. The search prompt includes a dialog summarization instruction configured to instruct a generative artificial intelligence model to generate and output a dialog summary. The search prompt is sent to a first generative model. In response to the search prompt, a search query is generated and output by the first generative model based on the dialog summary. The search query is sent to a search system. Search result data is determined based on an execution of the search query by the search system. At least some of the search result data is included in an output portion of the online dialog. The output portion is configured to be displayed at the computing device in response to the input portion of the online dialog.

Abstract: The present disclosure describes systems and methods relating to full motion video (FMV) routing in one-way transfer (OWT) systems. The present technology enriches the datagrams of the video stream that are sent from the low-trust side of the OWT system with a global unique identifier (GUID) that is used as an identifier to determine a particular destination on the high-trust side of the OWT system. The enriched video stream is then transmitted through an OWT system that provide high reliability for the enriched video stream. When the enriched video stream is received on the high-trust side, the GUID in the datagram is extracted and used to identify destination addresses for destination devices in the high-trust computing environment. The video stream is then delivered to the destination devices having the corresponding destination addresses.

Abstract: Examples of the present disclosure describe systems and methods relating to full motion video (FMV) routing in one-way transfer (OWT) systems. The present technology reserves a particular channel for transmission of a video stream, and then transmits the video stream from a low-trust computing environment to a high-trust computing environment along a data path defined by the channel. When the video stream is received on the high-trust side, the channel, on which the video stream is received, is determined and used to query a routing table that returns destination addresses of destination devices to which the video stream is to be transmitted. The video stream is then delivered to the destination devices having the corresponding addresses.

Abstract: This document relates to devices employing imaging devices, such as cameras and improved camera performance. In one example the device includes an optical element and a sensing element configured to sense light passing through the optical element. This example includes a set of MEMS actuators configured to be individually selectively controlled to create six degrees of freedom (6DoF) movement between the sensing element and the optical element.

Abstract: Systems and methods for generating dynamic quick actions for an application in a web browser. The dynamic quick actions correspond to various functions of an application accessible via a web browser sidebar interface. When a hover event is detected in association with an icon of the application, a quick-actions card is generated that includes quick actions of the application from which the user can select. For instance, a selection of a quick action triggers the web browser to execute an action that causes the application function to be performed. Thus, application functions are able to be surfaced and controlled via a single input device selection (e.g., a mouse click).

Abstract: Systems and methods for generating advanced feedback for draft messages using a language model are disclosed. Prior messages, along with corresponding reactions, may be incorporated into an AI prompt that is processed by a language model to generate an output payload. The output payload is processed to generate custom profiles for users that have provided the reactions to the messages. At runtime, while a draft message is being composed within a messaging application, the data from the draft message (and message thread where applicable) are received. The custom profiles for recipients of the draft message are then retrieved from the database of custom profiles. The data from the draft message as well as the retrieved custom profiles are incorporated into another AI prompt that is processed by the language model to produce another output payload. The output payload is post-processed to extract advanced feedback for the draft message.

Abstract: This document relates to personalized image or video processing. For example, the disclosed implementations can identify a designated user of a computing device that participates in a video call with other users. When another person appears in a video feed captured by the computing device, the other person can be removed. This can avoid distractions that can be caused, for example, by family members or pets that inadvertently walk into the field of view while a designated user is participating in a video call. Similar techniques can be employed to remove people other than designated users from still images.

Abstract: This document relates to distributed teleconferencing. Some implementations can employ adaptive audio or video enhancement to address scenarios where audio enhancement can tend to remove desirable sounds. For instance, adaptive audio enhancement can involve detecting the presence of a sound, such as clapping, and modifying audio enhancement so that the sound is retained in an enhanced audio signal. Adaptive video processing can involve detecting the presence of the sound and adding a graphical identifier to a video signal that conveys the presence of that sound.

Abstract: This document relates to active speaker detection using distributed devices. For example, the disclosed implementations can employ personal devices of one or more users to detect when those users are speaking during a call with other users. Then, a camera on the personal device can be employed to obtain a front-facing view of the user, which can be provided to other call participants. In some cases, a microphone and/or camera on the user's device are employed to detect when the user is actively speaking.

Abstract: The description relates to hinged devices. One example can include a flexible display extending over a major surface of a first portion across a hinge assembly and over a major surface of a second portion. The hinge assembly can include a pair of hinge sub-assemblies that include elongate slider links that extend along a length from proximal ends that are rotationally secured to the hinge assembly to distal ends that are slideably secured to the first portion. The hinge assembly can also include an auxiliary hinge sub-assembly interposed between the pair of hinge sub-assemblies and comprising a truncated elongate slider link that extends along a length that is less than the length of the elongate slider links and extends from a proximal end that is rotationally secured to the hinge assembly to the distal end that is slideably secured in the first portion.

Abstract: Described herein is a technique for generating data for a computer system. An apparatus comprises a high-speed communication bus, a memory unit communicatively coupled to the high-speed communication bus, an integrated circuit communicatively coupled to the high-speed communication bus, the integrated circuit to execute a machine learned model trained to receive input data and generate a set of implicit keywords from the input data to support a network service, and an implicit facet mapper to map the input data to an implicit facet of a knowledge graph, a database communicatively coupled to the high-speed communication bus, the database to store the set of implicit keywords and implicit facet in a data structure, and a network interface communicatively coupled to the high-speed communication bus to access a wireless network. Other embodiments are described and claimed.

Abstract: A data processing system implements receiving a request for an access token from a web-based nested application hosted by a native application, creating a secure authentication channel between the first web-based nested application and the native application that bypasses one or more intermediary nested applications to prevent the one or more intermediary nested applications from accessing data exchanged between the first web-based nested application and the native application; sending the request for the access token to the native application via the secure authentication channel; obtaining the access token from a token service on a remote server over a network connection; providing the access token to the first web-based nested application via the secure authentication channel; sending a request to access the content on the resource server and the access token to the resource server; and performing one or more actions on the content on the resource server.

Abstract: Innovations in packet scheduling, which allow a scheduling mechanism to approximate fine-grained rate adjustments and cycle-granularity inter-packet spacing for packets of a flow, are described herein. For example, in an iteration of a scheduler loop, a sender determines whether a proximity condition is satisfied for the next packet of a flow. The proximity condition depends at least in part on how long a target next send time is after a current time. The next packet is scheduled for transmission if the next packet is due or if the proximity condition is satisfied for the next packet. When the next packet is scheduled for transmission, the sender sends the next packet and updates the target next send time based at least in part on a target transmission rate.

Abstract: Systems and methods are provided for implementing content enrichment of document data and data source connector content. A computing system extracts enrichment data associated with a data item and ingests the extracted enrichment data within a data store, by indexing the extracted enrichment data in a search index of the data store. Indexing the extracted enrichment data includes generating a listing within the search index, the listing being generated to be searchable and refinable using a search engine, the extracted enrichment data being retrievable via the search engine. At least a portion of the enrichment data is generated using a large language model. The enrichment data may be stored in an enrichment data field in a schema file of the data item, in some cases, after the schema file has been modified to add the enrichment data field if not already contained in the schema file.

Abstract: Techniques are described for decoding frame masks in a video stream. In some solutions, a computer implemented method receives a frame of video data comprising an array of pixels with encoded color values defined in a YUV color space. The encoded values, converted from original color values using a mapping function, are bounded within a second data range, shifted or compressed from a first data range of the original color values. The method decodes the encoded color values to obtain decoded color values for the array of pixels and extracts a foreground mask based on the decoded color values. The foreground mask identifies a subset of the array of pixels to be in a foreground of the frame. The method then converts the decoded color values of the subset of the array of pixels to new color values by applying an inverse mapping function.

Abstract: An online meeting system begins the process of connecting a user to an online meeting when a request is received from a user to start or join the online meeting. In response to the request, a pre-meeting user interface (UI) is displayed having a user interface control which, when activated, enables transmission of audio and/or video data for the online meeting, and prior to activation of the control, connection parameters are shared with an online meeting service to preheat the meeting. In response to activation of the control, transmission of audio and/or video data for the online meeting is enabled.

Abstract: The methods and systems improve uncorrectable error (UE) and silent data corruption (SDC) rates for memory chips and improve error correction of the memory chips. The systems may include a memory bank with a plurality of memory chips in communication with a memory controller. The memory bank may use one additional memory chip that stores a bitwise parity of the data stored in the remaining memory chips of the memory bank. The parity bits are used to rebuild corrupted data when a UE occurs. The parity bits are also used to detect whether a SDC occurred in the data.

Abstract: Described herein is a system for calculating trust of a client session. A proof of work value is received from a session of a client computer. The proof of work value is calculated by the session of the client computer based, at least in part, upon a work function and input(s) received from service(s) connected to the session. A probability that the session is trustworthy is calculated based, at least in part, upon the proof of work value. Feedback is provided to the session of the client computer based, at least in part, upon the calculated probability. The feedback can increase complexity or frequency of calculation. The feedback can include an update to the work function.

Abstract: A system and method for automatically generating a workplan schedule for a team of employees in an organization includes receiving a request to create a workplan schedule for the team, the team including two or more team members who are employees of the organization and the workplan schedule identifying a work location for the team members. One or more collaborators for one or more of the team members is detected, where the collaborator is an employee of the organization who is not a member of the team and with whom at least one of the team members collaborate.

Abstract: Systems and methods are provided for maintaining a live state of a virtual function during VM fast save and fast restore operations. A computing system performs the fast save operation, by disabling network optimizations, while maintaining allocation and assignment of a VF to a VM and maintaining a VPCI bus connection between a VF device and hardware resources of a host device. The VPCI VSP is caused to save a VF device state to a runtime repository, while VF information is also saved to the runtime repository. A VM switch saves a synthetic network adapter, while a VM synthetic network adapter is torn down. The computing system subsequently performs the fast restore operation, by initializing the VM synthetic network adapter, retrieving the VF information from the runtime repository, causing creation of a VPCI bus interface, assigning the VF device, restoring the VF device state, and resuming the synthetic network adapter.

Abstract: A thermal management device includes a wicking heat spreader and a boiling enhancement surface feature positioned on at least one interior surface of the wicking heat spreader.

Abstract: Examples of the present disclosure describe systems and methods determining a root cause of an outage of a dependent service. A method includes detecting an outage of a dependent service, determining a first service dependency of the dependent service, and identifying one or more instances of the first service dependency by accessing a service provider of the first service dependency. The method also includes collecting one or more service level indicators (SLIs) for one or more instances of the first service dependency and determining a health status of the instances of the first service dependency using the SLIs. The method further includes determining a root cause for the outage of the dependent service based on the health status of the instances of the first service dependency.

Abstract: A computing system is provided, including a camera configured to capture a series of images in a video stream, processing circuitry, and an image processing pipeline including a preprocessing module and an encoder, in which the preprocessing module is configured to receive the video stream from the camera at a camera-native video data rate parameter. The processing circuitry is configured to communicate with a server to establish a joint video communication session with remote client devices and receive a requested video data rate parameter from the server for the video stream captured by the camera. The processing circuitry is further configured to send a command to the preprocessing module to adjust the camera-native video data rate parameter of the video stream to the requested video data rate parameter and preprocess the video stream at the requested video data rate parameter and pass the preprocessed video stream to the encoder.

Abstract: The technology described herein generates cryptographic keys for a system on-chip (SoC). The technology uses a cryptographic key hierarchy to generate cryptographic keys as needed and in constant time. The cryptographic key hierarchy uses a cryptographic key from a parent node as an input to the cryptographic function used to generate cryptographic keys associated with child nodes of the parent node. A single cryptographic key or batch of cryptographic keys may be generated as needed in response to a trigger event. Each key may be generated using a prescribed amount of iterations performed across the various hierarchical levels of the cryptographic key hierarchy. The amount of iterations performed at each level may be determined on a chip-by-chip basis. The unique identifier on a SoC may also be used as an input to the cryptographic function. Using the unique identifier as an input to the cryptographic function generates SoC-specific cryptographic keys.

Abstract: Systems and methods for providing an artificial intelligence (AI)-based interface for an application include receiving a prompt from a user interface (UI) component of an interface client that defines at least one task to be performed in the application. The prompt is supplied to at least one language model as input. The at least one language model is trained to process the prompt to identify the at least one task to be performed, generate new content if required by the at least one task, and domain-specific instructions for causing the tasks to be performed in the notes application. Notes domain-specific language (NDSL) instructions are provided as output to the notes application where they are executed in the notes application to perform the at least one task.

Abstract: Systems and methods for generating output data based on a data graph are provided. An output request for the output data based on the data graph is received. The output request comprises one of a natural language request from a target user or an application request from an application of the target user. The data graph has nodes and edges between the nodes. The nodes represent entities associated with an enterprise organization and the edges represent relationships among the entities. A graph data query is generated with a large language model (LLM) using the output request as a first input to the LLM. The graph data query is performed against the data graph to obtain a graph data output that represents a sub-portion of the data graph. The output data is generated with the LLM using the graph data output as a second input to the LLM.

Abstract: Sampling operations enable a computer vision tool to regulate downstream tasks. The sampling operations can indicate which frames of a video sequence should be processed by different downstream tasks. For example, a computer vision tool receives encoded data for a given frame and uses the encoded data to determine inputs for machine learning models in different channels. The computer vision tool provides the inputs to the machine learning models, respectively, and fuses results from the machine learning models. In this way, the computer vision tool determines a set of event indicators for the given frame. Based at least in part on the event indicator(s) for the given frame, the computer vision tool regulates downstream tasks for the given frame (e.g., selectively performing or skipping downstream tasks for the given frame, or otherwise adjusting how and when downstream tasks are performed for the given frame).

Abstract: A logic circuit includes an input data line, and a zero-detection element configured to output a latch control signal with a first state based at least in part on detecting that a current input value on the input data line is equal to zero. A latch is configured to receive the current input value and output a latch output value, wherein the latch output value is a prior input value based at least in part on the latch control signal having the first state, and wherein the latch output value is the current input value based at least in part on the latch control signal having a second state. A multiplier performs a multiplication operation based at least in part on the latch output value.

Abstract: A passive communication system includes a first reflector and a modulator unit. The first reflector is disposed at a first location within line of sight of a first satellite. The first satellite is configured to transmit a first signal at a first wavelength. The first reflector includes a reflective surface that reflects at least a portion of the first signal which is incident on the reflective surface back toward the first satellite, and the first satellite includes a detector for measuring reflected signals received at the first satellite. The modulator unit is configured to modulate a reflectivity of the reflective surface of the first reflector between a first reflective state to a second reflective state to adjust the portion of the first signal which is incident on the reflective surface that is reflected back toward the first satellite.

Abstract: A method and system are provided for processing natural language user queries for commanding a user interface to perform functions. Individual user queries are classified in accordance with the types of functions and a plurality of user queries may be related to define a particular command. To assist with classification, a query type for each user query is determined where the query type is one of a functional query requesting a particular new command to perform a particular type of function, an entity query relating to an entity associated with the particular new command having the particular type of function and a clarification query responding to a clarification question posed to clarify a prior user query having the particular type of function. Functional queries may be processed using a plurality of natural language processing techniques and scores from each technique combined to determine which type of function is commanded.

Abstract: Operators managing a cloud RAN collect vast amounts of data, e.g., node-level data, gNodeB level data, user level data, and flow-level data, which are utilized for network monitoring, evaluating key performance indicators (KPIs), and nodes management. Retrieving and visualizing information and values of different metrics is critical to managing network operation; however, data retrieval on large datasets is challenging. While foundation models perform poorly on large datasets, an accurate answer to a data query is generated by providing semantically similar metrics as context to a foundation model, thereby limiting the number of counters needed for processing the data query. The foundation model then generates a first output of metrics relevant to answering the data query and, based on the first output, generates a second output comprising query code (e.g., SQL or KQL) for computing the answer, thereby improving mathematical accuracy of the answer.

Abstract: As part of managing delivery of a given packet flow according to a reliable transport protocol, a sender sends, to a receiver, a last flow packet among multiple flow packets of a flowlet. After sending the last flow packet but before satisfaction of a timeout condition for the last flow packet, the sender sends one or more end-of-flowlet (“EOF”) packets, which can be flush packets, query packets, or another type of packet. The sender receives, from the receiver, feedback metadata for the EOF packet(s) and updates a tracking window based at least in part on the feedback metadata. The sender selectively resends one or more unacknowledged flow packets according to the updated tracking window. In this way, the sender can quickly address any dropped packets or significantly delayed packets at the end of a flowlet, without waiting for the timeout condition to detect the dropped or delayed packets.