Abstract: Techniques for improving the power consumption of a device without impacting or with minimal impact to operations of the device are described. In an example, the device includes a processor. The device receives, while the processor is operating in a first power mode, first input data corresponding to first audio detected by a microphone. Based at least in part on the first input data, the device detects a sound event or ambient noise. Based at least in part on a detection of the ambient noise only, the device causes processor to operate in a second power mode in which the processor consumes less power than in the first power mode.

Abstract: Techniques for classifying and processing physical objects are disclosed. In an example, a computer system may receive first data indicating that a first machine learning model of a robotic system is incapable to classify a physical object according to at least one of a set of predetermined classifications. The computer system may also receive second data corresponding to one or more attributes associated with the physical object. A second machine learning model of the computer system may determine a cluster of physical objects that includes an identifier of the physical object, whereby the identifier is included in the cluster based at least in part on the first data and a common attribute with other physical objects of the cluster. The computer system may then determine data for processing subsequent physical objects that are determined to have the common attribute.

Abstract: Devices and techniques are generally described for generating response templates for natural language processing. In various examples, a first knowledge graph comprising a plurality of entities may be received. First text data may be received for a first response template, the first text data defining a natural language input configured to invoke the first response template. A response definition may be received for the first response template, the response definition defining a response associated with the first response template. Natural language input data may be received. A determination may be made that the natural language input data corresponds to the natural language input configured to invoke the first response template. The first response template may be configured to generate natural language output data.

Abstract: Data clean rooms provided for subscribers to advertising analysis services include data regarding advertising impressions, selections, view or conversion events, and memberships of audience segments, and computing functions or capacities for processing the data in an anonymized fashion. When advertisements are presented to customers, and interactions by customers (e.g., impressions, views, clicks, conversions or others) with goods or services are detected, data regarding the interactions may be aggregated and stored. When queries on behalf of an advertiser are identified, data associated with the queries may be retrieved and stored in a data clean room for processing or analysis. Anonymized responses to the query may be returned to the customer or serve as basis for one or more advertising events.

Abstract: Systems and processes are described for a message service with distributed key caching for server-side encryption. Message requests are received by message handlers of the message service that cache data encryption keys used to encrypt and decrypt messages that are stored to message containers in back end storage. A metadata service obtains the data encryption keys from a key management service, caches the keys locally, and sends the keys to the message handlers upon request, where the keys are cached, again. The key management service may generate the data encryption keys based on a master key (e.g., a client's master key). The message handlers may send both message data encrypted using the data encryption key and an encrypted copy of the data encryption key to be stored together in the data store.

Abstract: Disclosed are systems and methods that determine specification portions of Dafny code and transform those specifications into one or more annotations, expressions, comments, and/or assertions that are included in a destination code written in a destination language as part of a compilation of the Dafny code into the destination code. The annotations, expressions, comments, and/or assertions in the destination code may be utilized by a verification component, such as a Checker Framework, to detect errors that are introduced into the destination code by the compiler as part of the compilation or to verify the absence of errors in the destination code.

Abstract: Determining whether to allow access to a message is disclosed. A message is received from a sender. The message is associated with a first time-to-live (TTL) value. A determination is made that the first time-to-live value has not been exceeded. The determination is made at least in part by obtaining an external master clock time. In response to the determination, access is allowed to the message.

Abstract: Joint noise and echo suppression may be performed for enhancing two-way audio communications. Audio data is captured at a communication device and audio data transmitted to the communication device from another communication device are used as input features to a trained machine learning model that uses the transmitted audio data as a reference signal to eliminate residual echo in the captured audio data when also suppressing noise in the captured audio data.

Abstract: Techniques for establishing a data connection are described. In an example, a computer system receives, from a second device of a computer network, first data associated with a first device and second data associated with the second device. The first device is not connected to the computer network. The computer system determines third data generated by one or more devices other than the first device and the second device and associated with at least one of: the first device, the second device, a user account, or the computer network. The computer system generates, based on the first data, the second data, and the third data, a confidence score indicating a likelihood of a user authorization to connect the first device to the computer network. The computer system sends, to the second device based on the confidence score, instructions associated with connecting the first device to the computer network.

Abstract: Systems, methods, and computer-readable media are disclosed for flexible rail switching components for container shuttles. In one embodiment, an example system for a shuttle may include an elongated spring steel plate, a first rail track component coupled to the elongated spring steel plate, a second rail track component coupled to the elongated spring steel plate and disposed adjacent to the first rail track component, and a third rail track component coupled to the elongated spring steel plate and disposed adjacent to the second rail track component. The elongated spring steel plate may be configured to be actuated from a first position to a second position. A distance between the first rail track component and the second rail track component when the elongated spring steel plate is in the second position may be less than when the elongated spring steel plate is in the first position.

Abstract: A constellation of many satellites provide communication between devices such as user terminals (UTs) and ground stations that are connected to other networks, such as the Internet. A constellation management system (CMS) facilitates management and operation of the satellites in the constellation and facilitates information exchange with other authorized systems to provide for situationally aware operation. The CMS may ingest data such as satellite telemetry, space weather data, object ephemeris data about other orbital objects, and so forth. The CMS uses the ingested data to automatically operate satellites to perform routine activities such as station keeping maneuvers, maintenance activities, interference mitigation, and so forth. Confirmation from a human operator may be obtained before performing some activities. Activities may be planned and coordinated to minimize resource consumption for the individual satellite as well as the constellation.

Abstract: An assembly includes a frame and a conveyor mat configured to transition between a retracted state in which the conveyor mat is disposed within a footprint of the frame, and a deployed state in which the conveyor mat is disposed at least partially disposed outside the footprint of the frame. The conveyor mat has a plurality of linkages, a plurality of rollers, a plurality of supports, and a plurality of axles. Individual axles couple to individual linkages of the plurality of linkages, individual rollers of the plurality of rollers, and individual supports of the plurality of supports together. A tensioning system is configured to control the conveyor mat transitioning between the retracted state and the deployed state.

Abstract: Disclosed are various embodiments for efficiently identifying whether any expected user interface elements (e.g., informational components, containers, navigational components, input controls, labels, etc.) are missing from a rendered user interface. A user interface is generated and rendered by a client device according to user interface content data received from a content provider. User interface feature identifiers associated with user interface features included in the rendered user interface are identified and applied to one or more probabilistic data structure algorithms to generate a rendered feature probabilistic data structure digest representing the rendered user interface features. The rendered feature probabilistic data structure digest is compared with an expected feature digest to determine whether any expected user interface features are missing from the rendered user interface.

Abstract: Techniques for monitoring devices to use ultrasonic signals to detect and track the locations of moving objects in an environment. To determine distance information, the monitoring devices emit a frequency-modulated continuous wave (FMCW) signal at an ultrasound frequency range. Reflections of the FMCW ultrasonic signal are used to generate time-of-arrival (TOA) profiles that indicate distances between the monitoring device and objects in the environment. The reflections can be processed to suppress undesirable interferences, such as reflections off non-mobile objects in the environment (e.g., walls, furniture, etc.), vibrations off the floorings or the ceilings, etc. After processing the reflections, a heatmap can be used to plot the intensity of the reflections for the different TOAs of the reflections, and depict the movement of the user over time. Finally, a Kalman filter is used to smooth the peaks in the intensity values on the plot, and determine the trajectory of the human.

Abstract: A satellite provides communication between user terminals (UTs) and ground stations that connect to other networks, such as the Internet. Because the satellite is within range of many UTs at any given time, many UTs are in contention to use an uplink to send upstream data to the satellite. This saturates a random-access channel (RACH) on the uplink. When a UT has data to uplink, it sends a short buffer data status (SBDS) message using the RACH. The minimal size of the SBDS facilitates use of a non-orthogonal multiple access uplink. Based on the SBDS, the satellite allocates a grant to the UT to use the uplink. Additional messages from the UT involving buffer status may be sent using the granted uplink. Unsolicited grants may be issued to the UT based on analysis of uplink and downlink traffic. If needed, the RACH may still be used to request additional grants.

Abstract: At a first compute instance run on a virtualization host, a local instance scaling manager is launched. The scaling manager determines, based on metrics collected at the host, that a triggering condition for redistributing one or more types of resources of the first compute instance has been met. The scaling manager causes virtualization management components to allocate a subset of the first compute instance's resources to a second compute instance at the host.

Abstract: A method and apparatus for testing and simulating an access control policy are disclosed. Evaluating an access control policy may be performed by utilizing a deny statement that causes the access request to be rejected despite actions indicated in the access request being authorized. Further, an independent simulation environment may be utilized for testing access control policy evaluation.

Abstract: A hardware security module (HSM) client processes a request to store data in a set of HSMs. The HSM client determines a property of the data indicative of a sensitivity classification of the data. As a result of determining the data lacks a classification as sensitive, the HSM client transmits the data to a data store outside the set of HSMs and updates a database used by the HSM client to associate an identifier of the data with a reference to a location in the data store.

Abstract: Systems, methods, and computer-readable media are disclosed for robotic system camera calibration and localization using robot-mounted registered patterns. In one embodiment, an example robotic system may include a robotic manipulator, and a picking assembly coupled to the robotic manipulator, where the picking assembly is configured to grasp and release items, and where the picking assembly includes a housing having a first flat surface. Example robotic systems may include a first calibration pattern disposed on the first flat surface of the housing, a first camera configured to image the first calibration pattern, and a controller configured to calibrate the robotic system.

Abstract: A speech recognition platform configured to receive an audio signal that includes speech from a user and perform automatic speech recognition (ASR) on the audio signal to identify ASR results. The platform may identify: (i) a domain of a voice command within the speech based on the ASR results and based on context information associated with the speech or the user, and (ii) an intent of the voice command. In response to identifying the intent, the platform may perform multiple actions corresponding to this intent. The platform may select a target action to perform, and may engage in a back-and-forth dialog to obtain information for completing the target action. The action may include streaming audio to the device, setting a reminder for the user, purchasing an item on behalf of the user, making a reservation for the user or launching an application for the user.

Abstract: A multi-channel transformer acoustic model that processes a plurality of audio signals output by microphones of a microphone array and outputs probabilities for acoustic units of an utterance represented in the audio signals. The audio signals represent the individual microphones' respective capturing of the utterance. The multi-channel model may perform self-attention on embeddings of the audio signals and then cross-channel attention across the attended audio signals. The cross-channel attention may involve processing of signals relative to each other to model the relationships across channels within and across time frames. The multi-channel model may include a transducer to perform processing frame-by-frame.

Abstract: A system configured to improve track selection while performing audio type detection using sound source localization (SSL) data is provided. A device processes audio data representing sounds from multiple sound sources to determine SSL data that distinguishes between each of the sound sources. The system detects an acoustic event and performs SSL track selection to select the sound source that corresponds to the acoustic event based on input features. To improve SSL track selection, the system detects current conditions of the environment and determines adaptive weight values that vary based on the current conditions, such as a noise level of the environment, whether playback is detected, whether the device is located near one or more walls, etc. By adjusting the adaptive weight values, the system improves an accuracy of the SSL track selection by prioritizing the input features that are most predictive during the current conditions.

Abstract: Systems and methods for account data association with voice interface devices are disclosed. For example, when a host user/primary user and guest user have consented for account data to be associated with the primary user's devices, a request to associate the account data may be received. Voice and device-based authentication may be performed to confirm the identity of the guest user and the guest user's account data may be associated with the primary user's devices. During a guest session, voice recognition may be utilized to determine if a given user utterance is from the guest user or the primary user, and actions may be performed by the voice interface device accordingly.

Abstract: Communication in an asymmetric multiengine system is handled using engine routing tables defining subsets of engines to control engine-to-engine connection mapping. Local devices perform an engine selection process that includes selecting an engine routing table based on a number of remote engines in a remote device and selecting an engine set from the selected table based on an identifier of the remote device. A connection to the remote device is created using the engines identified in the selected engine set.

Abstract: Systems and methods are provided for prioritized restoration of multiple network connections that have been interrupted. Connection state data (e.g., quantity of pending IO requests) may be used to prioritize reestablishment of some connections over others. In addition, the connections may be reestablished using “in-band” handshakes over the previously-established network connections.

Abstract: Techniques for managing latency of communications between compute instances and mobile devices are described. A message including an indication of a mobility event associated with a mobile device of a communications service provider network is received. The mobility event indicates a change in a connection point of the mobile device to the communications service provider network from a from a first access point to a second access point. A communications delay of at least a portion of a network path between the mobile device and a compute instance via the second access point is determined to not satisfy a latency constraint. A second provider substrate extension of the cloud provider network that satisfies the latency constraint for communications with the mobile device via the second access point is identified, and a message is sent to the second provider substrate extension to cause the launch of another compute instance.

Abstract: Embodiments herein describe a pick or stow system that uses a mechanical system to stow an item into a container, retrieve an item from the container, or both. In one embodiment, the mechanical system includes a gantry or a robotic arm that permits various sensors, tools, or end effectors to interact with a plurality of containers. For example, the plurality of containers may be supported by a rack where each container may have any number of types of items. The gantry or robotic arm can move the sensors, tools, or end effectors to a particular container to either stow an item into the container or pick an item from the container.

Abstract: Visual content of media programs is recognized using text-based scripts of the media programs. A script of a media program includes sets of words to be spoken by actors during the media program, and also descriptions of points of interest of the media program. A file of subtitles or captions of a media program includes sets of words actually spoken by actors during the media program along with marks or stamps of times when such words were spoken. The sets of words of a script and in subtitles or captions are processed to determine where such words align. The sets of words of the script are then marked or stamped with times from the subtitles. The descriptions may then be processed to predict the visual content from such descriptions, and times at which the visual content appears within the media program are determined from the sets of words.

Abstract: Techniques are described for enabling a cloud-based service to provide rate controls for events to be delivered by the service to event targets. An event bus service, for example, broadly enables users to create and configure event buses that receive events from event sources and from which the event bus service filters, transforms, routes, and delivers events to selected event targets according to configurable rules. A new event egress call pattern is provided that estimates a rate of attempted event traffic to event targets and a rate of successfully delivered event traffic (also referred to as a local rate estimation), and that uses the rate estimations to determine when it may be appropriate to throttle event delivery, thereby reducing stress on downstream systems experiencing capacity issues and potentially reducing a number of event delivery retries to be handled by the event bus service.

Abstract: Devices, systems, and methods are provided for generating and selecting video clips for inclusion in video sequences based on still frame images. A method may include encoding first embeddings for a first video including first images of an item at a first scene, the first embeddings indicative of features of the first scene; encoding second embeddings for a second video including second images of the item at a second scene, the second embeddings indicative of features of the second scene; encoding third embeddings for the first video, the third embeddings indicative of features of a first type of camera shot used for the first images; encoding fourth embeddings for the second video, the fourth embeddings indicative of features of a second type of camera shot used for the second images; and generating, based on the first, second, third, and fourth embeddings, a video sequence for the item.

Abstract: During an episode of a media program, a creator of the media program requests that listeners join and participate in the episode. In reply, one or more listeners provide audio data to a control system of the media program, e.g., by speaking utterances that are captured by devices of the listeners. The audio data is processed to identify the users, and to determine attributes of the audio data, to identify words expressed in the audio data, and to determine features such as sentiments or opinions of the audio data. The listeners that provided the data are ranked or scored based on information regarding the respective listeners, the attributes of the audio data, or the words or sentiments. One or more of the listeners are recommended to the creator to be permitted to participate in the episode of the media program, or automatically joined in the episode.

Abstract: Aspects of a data environment, such as the cloning, hibernation, and recovery of databases, are managed using a separate control environment. A monitoring component of the control environment can periodically communicate with the data environment to determine any necessary actions to be performed, such as to recover from faults or events for a data instance in the data environment. A workflow can be instantiated that includes tasks necessary to perform actions such as recovery, hibernation, resumption from hibernation, or backup or cloning. Tasks of the workflow can cause certain jobs to be performed by host managers in the data environment to affect calls made to the control environment.

Abstract: At a radio-based application pipeline processing server at which a portion of a distributed unit (DU) of a radio-based application is implemented, a particular networking hardware device is selected from among several devices (which include least one device incorporated within a network function accelerator card and at least one device which is not part of an accelerator card) for transmission of at least a portion of mid-haul traffic to a centralized unit (CU). The mid-haul traffic is transmitted to the CU via the selected device. At least a portion of front-haul traffic is transmitted to a radio unit (RU) via a networking hardware device incorporated within a network function accelerator card of the server.

Abstract: Techniques for computer system-based conversations are described. In an example, a system receives, from a first device, first data corresponding to a first interaction in a conversation that requests a function. The system causes the first device to output a first response to the first interaction. Prior to an execution of the function, the system determines that the conversation is to be paused and causes the first device to output a first indication that the conversation is paused. Upon determining that the conversation is to be resumed, the system causes a second device to output a second indication that the conversation is resumed. The second device can be the same or different from the first device. The system receives, from the second device, second data corresponding to a second interaction in the conversation and causes the execution of the function based at least in part on the second data.

Abstract: A text-to-speech (TTS) system may be configured to imitate characteristics of a target voice based on a limited dataset. The TTS system may include a machine learning model pre-trained using a synthetic parallel dataset and fine-tuned using examples of the target voice. A TTS component trained using a large single-speaker dataset may be used to generate the synthetic parallel dataset based on a multi-speaker dataset. The synthetic parallel dataset may include target audio data representing speech in the multi-speaker dataset and predicted audio data generated by the TTS component based on transcripts of the speech. The machine learning model may be pre-trained using the synthetic parallel dataset and fine-tuned using audio data representing target voice speech and predicted audio generated by the TTS component based on transcripts of the target voice speech. The trained model may be used to modify synthetic speech to approximate the characteristics of the target speech.

Abstract: An autonomous mobile device (AMD) moves in a physical space and uses a dock located at a first pose. Sometimes a user may move the dock. To determine whether the dock has moved, the AMD compares first pose data that was previously obtained while at the dock in the first pose to second pose data, such as after a restart or after leaving the dock. If a difference between the first pose and the second pose is greater than a threshold, the dock has moved. The pose data may be determined using a simultaneous localization and mapping (SLAM) algorithm to process images from cameras on the AMD. Transform data that relates coordinates of the second pose to coordinates of the first pose may be determined. If the dock is used as an origin for a map of the physical space, the transform data may be used to update the map.

Abstract: Techniques for compiling firewall rules into byte code or assembly code that can be loaded into cache memory of a processor and executed to evaluate received data packets. Rather than representing firewall rules in mid- or high-level languages stored in main memory, the techniques described herein include compiling the firewall rules into bytecode or assembly code, and distributing the code to the data plane. A packet-processing device may load the code representing the firewall rules into instruction cache of the processor. Further, the packet-processing device receives a data packet and extracts packet context data indicating attributes of the packet, and load the packet context data into a data cache of the processor. The processor can then execute the byte code or assembly code representing the firewall rules to evaluate the packet context data without having to access main memory to determine whether allow or block the data packet.

Abstract: A scalable encoding architecture is able to support multiple encoder versions without requiring new versions of the architecture to be generated and supported for each encoder version. An encoding profile template can be generated that includes the available settings for an encoder. A versioned schema can be generated that defines the settings for a particular encoder version. The schema can be persisted such that when a new channel or media stream is to be generated, the schema can be loaded and its values validated in order to cause the encoder to encode the media content using the appropriate values. An interface presented to the user to generate the template or channel can be dynamically generated such that only the applicable and/or relevant settings for a particular encoder version are displayed.

Abstract: A network-connected device service receives a request to authenticate a network-connected device. The network-connected device service determines, from a digital certificate identified in the request, a set of parameters of the digital certificate. The network-connected device service utilizes the set of parameters to identify, from a set of digital certificate clusters, a digital certificate cluster associated with the set of parameters. Through an audit of the digital certificate clusters, the network-connected device service determines whether the digital certificate cluster is indicative of the digital certificate being anomalous.

Abstract: An agent is installed on a computing device of a vehicle (e.g., an ECU); the agent receives messages from an application installed on the same computing device. The computing device is connected to a gateway computing device via a physical communication channel (e.g., CAN bus or Ethernet cable). When the gateway agent receives a message from the application, it converts the message into a different protocol based on a mapping of the protocols to physical communication channels. The message is sent to a gateway agent at the gateway device, where it is converted back to the original protocol so the message can be processed/transmitted to the internet. By using an agent, the in-vehicle application does not need to know about the vehicle's network architecture or changes to it in order to access the internet.

Abstract: A backbone service exposes the efficiency of the backbone network using tunnels between any source-destination pairs. The efficiency can be calculated using a ratio of a transmission rate of the packets through a tunnel and a number of hops in the tunnel. The efficiency also can be tracked based upon services using the backbone network in a compute service provider environment. The data generated by the backbone service can be transmitted, stored or displayed for informational purposes to provide insights to service teams on how to improve efficiency in the backbone by change routing behavior, such as by moving caches, adding or removing hops or changing advertisements to network devices.

Abstract: Respective correlation metrics between token groups of a particular text attribute of a data set and a prediction target attribute are computed. Based on the correlation metrics, a predictive token group list is created. For various observation records of the data set, values of a derived categorical attribute corresponding to the particular text attribute are determined based on matches between the particular text attribute value and the predictive token group list. A measure of the predictive utility of the particular text attribute is obtained using correlations between the categorical attribute and the prediction target attribute.

Abstract: Optimistic concurrency may be implemented for multiple writers to a database. Writes to a database from different writers may succeed unless a conflict is detected at the data store for the database. For detected conflicts, a request to resolve the conflict between writes may be submitted to a primary writer, which may determine and provide a resolution of the conflict to the conflicting writers. The data store may then be updated to commit the selected write identified by the resolution of the conflict.

Abstract: The disclosure describes, in part, techniques for attempting to diagnose the cause of a wireless connection problem in a client computing device and, in response, performing one or more actions for remedying the problem. In some instances, the techniques describe the client device communicating with one or more other client devices in the environment to determine whether they are connected to the WAP using the same password or a different password as the client device. If the former, then the client device may determine that the network stack is likely the cause of the connection error and may attempt to restart the network stack to current problem. If the other devices indicate that they are using the same password, however, then the client device may prompt a user to re-enter the network password.

Abstract: Systems and methods are provided to implement a database request router that routes queries to database servers storing database replicas. In embodiments, the request router maintains caching state metadata that indicates respective caching states of the database servers. For an incoming query, the request router selects a database server to receive the query based on the caching states of the database servers to improve cache utilization. In embodiments, the caching state metadata indicates whether different database objects used by observed queries are cached at individual servers. The database objects used may be determined from execution plans for the queries, obtained from the database servers. In embodiments, the request router may determine a querying pattern of the database and use it to predict an expected time for certain queries. The request router may generate a priming query ahead of the expected time to prepare database server cache for the expected queries.

Abstract: Aspects described herein relate to securely performing cross-boundary backup operations. A service of a computing resource service provider may enable backup operations between a source account and a destination account of an organization based at least in part on a security policy allowing such operations.

Abstract: Systems and methods are provided for using an algorithm and data structure for efficient and accurate classification of data items into recovery classes. When a target recovery time (TRT) is specified for a data set, a system may obtain version metadata regarding data items in the data set. The metadata may be obtained in reverse chronological order such that the latest record representing a version or other storage operation is first, followed by the second latest record, and so on. The system may use a bidirectional doubly linked list to efficiently store version data for a particular data item in memory. As version metadata records are read and added to the data structure in reverse chronological order, classification determinations may be triggered when certain conditions are met.