Abstract: Techniques for predicting a manufacturer and/or contents of a received package are described herein. Images of a package may be received from cameras. A visual vector for the package in a vector space of package attributes may be generated using the images. Physical attributes of the package may be received. A subset of candidate packages may be determined by comparing the physical attributes of a plurality of historically received packages to the physical attributes of the package. A ranking of the subset of candidate packages may be determined by identifying a distance in the vector space of the package attributes between a vector for the subset of candidate packages and the visual vector for the package. An identifier associated with a particular package of the ranked subset of candidate packages may be obtained. Data for the particular package may be retrieved from a database using the identifier.

Abstract: Techniques are generally described for onboarding optimization for new hires. In various examples, labor order data specifying a target number of new hires may be received. In some examples, a first machine learning model may predict a number of new hire appointments to offer based at least in part on the labor order data. In further examples, an optimization component may allocate the number of new hire appointments to respective time slots based at least in part on candidate preferences. In some additional examples, data may be sent to a first application effective to display a list of available new hire appointment time slots of the respective time slots.

Abstract: Systems, methods, and computer-readable media are disclosed for systems and methods for automated generation and presentation of sign language avatars for video content. Example methods may include determining, by one or more computer processors coupled to memory, a first segment of video content, the first segment including a first set of frames, first audio content, and first subtitle data, where the first subtitle data comprises a first word and a second word. Methods may include determining, using a first machine learning model, a first sign gesture associated with the first word, determining first motion data associated with the first sign gesture, and determining first facial expression data. Methods may include generating an avatar configured to perform the first sign gesture using the first motion data, where a facial expression of the avatar while performing the first sign gesture is based on the first facial expression data.

Abstract: Techniques are described for managing communications between multiple computing nodes, such as for computing nodes that are part of managed virtual computer networks provided on behalf of users or other entities. In some situations, one or more of the computing nodes of a managed virtual computer network is configured to perform actions to extend capabilities of the managed virtual computer network to other computing nodes that are not part of the managed virtual computer network, such as by forwarding communications between computing nodes of the managed virtual computer network and the other external computing nodes so as to enable the other external computing nodes to participate in the managed virtual computer network. In some situations, the computing nodes may include virtual machine nodes hosted on one or more physical computing machines or systems, such as by or on behalf of one or more users.

Abstract: Systems and methods are provided for evaluation of networks and changes thereto using automated analysis of network models. The automated analysis can be used to determine how to implement and mutate networks efficiently and effectively, to determine whether and why network resources are unable to communicate with each other, and the like. Automated analysis can allow users (e.g., network administrators) to define networks and pose changes to networks using high-level policies (e.g., written in a declarative language), have those polices automatically translated to lower-level implementation operations for analysis, and in some cases have results of the analysis presented back to the users in an easy-to-understand form.

Abstract: Techniques are described for maintaining contextual data to support content-related actions. In an example, a system stories second content at a source. The source is associated with first content. The system sends, to a device, an object that indicates the first content. From the device at a first time, the system receives first data indicating a first request for the second content and including source information that indicates the source. From the device at a second time, the system receives second data indicating a second request for the second content, the second data including the source information, the first data and the second data received at a frequency indicated by the object. The system determines that the requests are associated with the first content based on the source information included in the received data, and stores third data indicating a presentation of the first content by the device.

Abstract: Devices and techniques are generally described for regionalizing content based on the availability of similar content. In some examples, a first search query related to a first item may be received. A determination may be made that the first search query is associated with a first region. In some examples, the first item may be determined to be associated with the first region based at least in part on at least one of the first item or items classified as substitutes for the first item being available in a plurality of regions. A first plurality of search results associated with the first region may be determined. In further examples, code may be generated to cause the first computing device to display the first plurality of search results.

Abstract: The disclosure describes an automated aerial vehicle (AAV) and system for automatically detecting a contact or an imminent contact between a propeller of the AAV and an object (e.g., human, pet, or other animal). When a contact or an imminent contact is detected, a safety profile may be executed to reduce or avoid any potential harm to the object and/or the AAV. For example, if a contact with a propeller of the AAV by an object is detected, the rotation of the propeller may be stopped to avoid harming the object. Likewise, an object detection component may be used to detect an object that is nearing a propeller, stop the rotation of the propeller, and/or navigate the AAV away from the detected object.

Abstract: This disclosure describes, in part, techniques for modifying network applications. For instance, remote system(s) may store an application on a first virtual storage device and store a modification for the application on a second virtual storage device. The remote system(s) may then install the application on a virtual server using the first virtual storage device. Additionally, the remote system(s) may receive a request to access the application along with an indication to enable the modification for the application. Based on the request, the remote system(s) may modify the application on the virtual server using the second virtual storage device. The remote system(s) may then launch the application, as modified, on the virtual server. Once the application is launched, the system(s) may provide the application to a user device using the virtual server.

Abstract: A method and system including a monitoring device to detect, during a session, a set of radar signals associated with first respiratory-related movements of a first person and second respiratory-related movements of a second person located within a detection zone comprising at least a portion of a bed. The method and system dynamically generate an out-of-bed zone adjacent to a side of the bed. The out-of-bed zone is monitored to detect motion within the zone to determine when the first person exits and enters the bed. The method and system identify a portion of the set of radar signals collected during a time period when the first person is determined to be out of the bed. The identified portion of the set of radar signals is removed from the set of radar signals collected during the sleep session.

Abstract: Various embodiments of a system and method for generating a delivery plan for multi-tier delivery schemes are described. Embodiments may include generating a delivery plan for a region including multiple delivery locations. The delivery plan may specify at least one agent delivery route within the region. The delivery route may include at least one shipment acquisition location coinciding with an expected travel route of a vehicle conveying shipments. The agent delivery route may be traversed by an agent that is to deliver a specific group of shipments acquired from the vehicle at the shipment acquisition location. Embodiments may include selecting the specific group of shipments for delivery on the delivery route. The selecting is based on a time constraint for delivering the group of shipments to respective delivery locations along the delivery route, and historical delivery time information for previous deliveries to the delivery locations along the delivery route.

Abstract: Techniques are described for enabling model-driven server migration workflows in a cloud provider network. Cloud provider networks often provide various types of tools and services that enable users to migrate computing resources (e.g., servers, databases, applications, etc.) from users' on-premises computing environments to a cloud provider network. A model-driven server migration service as described herein comprises a plurality of modular migration components including, e.g., a snapshot validation component, a snapshot conversion component, an injection component, etc. The model-driven server migration service enables users to customize server migration workflows using server migration templates containing descriptive configurations for some or all of the provided migration components.

Abstract: Time and value ordering may be applied for items stored in data backups. A change log that persists changes to a data set may be updated with changes and used to update an in-memory table for the data set, which describes changes to items up to a current time. An event may be detected to seal the in-memory table from subsequent updates and a persistent data object that orders the items in the in-memory according to both keys of the respective items and the respective time values of the items, as stored in the change log, may be generated and stored as part of a backup for the data set.

Abstract: Modular storage systems include load sensors disposed beneath or in association with platforms. The load sensors are pulse-sampled for signals corresponding to loads at time intervals, rather than continuously, in order to conserve electrical power. An item associated with a transaction is identified based on a change in the loads, as determined from the pulse-sampled signals. The platforms are aligned at horizontal or non-horizontal angles, and photovoltaic power sources with banks of photovoltaic cells are mounted to front edges of the platforms. When a user reaches over one of the photovoltaic cells to access an item on a platform, a level of power or voltage generated by the photovoltaic cell is diminished. The item closest to the photovoltaic cell having the diminished power or voltage, or a customer that accessed the item, may be identified accordingly.

Abstract: A domain registry (DR) service executing within a service provider network protects data, such as account data, that is associated with different accounts for testing and/or performing other operations/actions by registering an account with one or more domains. The DR service may register an account in one or more domains based on a request by a user. The operations performed that uses/changes data associated with accounts may be restricted based on the domains for which an account is registered. For example, an account that is registered in a “testing” domain may have different workflows/operations performed using the account data associated with the account registered in the testing domain as compared to an account that is not part of the testing domain.

Abstract: A plurality of training iterations is conducted for a machine learning task. A given iteration includes generating a version of a stacking model using a portion of a labeled data set. Proposed labels are then obtained in the iteration using the generated version of the stacking model for a set of unlabeled records. The unlabeled records and their proposed labels are then used to generate versions of base models for the iteration. After the training iterations are completed, a trained ensemble of models including the stacking model and the base models is stored.

Abstract: A request is received associated with a review. Within first content, a first field of interest and a second field of interest are identified and within second content, a third field of interest and a fourth field of interest are identified. A review is generated that includes a first indication of the first field of interest and a second indication of the second field of interest within the first content, as well as a third indication of the third field of interest and a fourth indication of the fourth field of interest within the second content. The review is transmitted to a device of a reviewer for reviewing the content.

Abstract: Systems and methods for image-based item identification are disclosed. Image data corresponding to one or more images depicting an item may be sent to one or more remote systems for image-based item identification. The identification indications and/or identification confidence scores received from the remote systems may be aggregated and weighted based at least in part on one or more factors related to the remote systems, the results, domains, image capture timing, image capture angles, and/or events to more accurately identify an item depicted in the images.

Abstract: Modular components may be used to build a robotic manipulator. A subset of the modular components can be selected to build the robotic manipulator based on a schematic. The subset of modular components can be assembled in different combinations to build the robotic manipulator. Using one of the combinations of the subset of modular components, the robotic manipulator can be built.

Abstract: Techniques for a service provider network to communicatively couple services and/or applications in a serverless computing environment. A pipe component can configure a pipe to integrate two services by transmitting data between services and/or applications using the pipe. The pipe may also be configured to transform how a service processes an event, control timing of event transmissions using the pipe, define an event structure for an event, and/or batch events. Pipes enable an application or service to exchange data with a variety of services provided by the service provider network while controlling what type of data is generated, stored, or transmitted.

Abstract: Devices and techniques are generally described for detection of network anomalies. In various examples, first data describing network communication between a plurality of source entities and a plurality of destination entities may be received. In some examples, respective feature data representing network communication between a respective source entity and one or more of the plurality of destination entities may be generated. In some examples, an unsupervised machine learning model may be used to determine a first number of clusters of the feature data. In various cases, a first source entity that is an outlier with respect to the first number of clusters may be determined based at least in part on the first number of clusters. The first source entity may be classified as an anomalous entity.

Abstract: A system that implements a data storage service may store data for a database table in multiple replicated partitions on respective storage nodes. In response to a request to back up a table, the service may back up individual partitions of the table to a remote storage system independently and (in some cases) in parallel, and may update (or create) and store metadata about the table and its partitions on storage nodes of the data storage service and/or in the remote storage system. Backing up each partition may include exporting it from the database in which the table is stored, packaging and compressing the exported partition for upload, and uploading the exported, packaged, and compressed partition to the remote storage system. The remote storage system may be a key-value durable storage system in which each backed-up partition is accessible using its partition identifier as the key.

Abstract: A system for encoding video for film grain synthesis is disclosed. A video encoding system includes a film grain description estimator configured to receive an input video comprising film grain and analyze the input video to generate a film grain description for the input video. The video encoding system includes a video encoder configured to receive the input video and, responsive to a determination that the input video comprises film grain, configure one or more encoder settings based on presence of the film grain to achieve a compression target for the input video. The video encoder is further configured to encode the input video comprising film grain using the encoder settings configured based on the presence of film grain to produce encoded video. The video encoding system is configured to send the encoded video and the film grain description to a video destination.

Abstract: Techniques for performing centralized unsuperivised learning in a multi-domain system are described. A user may request labeled data for an ML task, where the request includes a prompt for obtaining relevant explicit user feedback. The system may use the prompt to collect explicit user feedback for relevant runtime user inputs. After a duration of time (in the user's request for labeled data) has elapsed, the system determines whether collected user feedback indicates processing of the user input was defective and, if so, determines a cause of the defective processing. The system then uses one or more label generators to generate labeled data using the collected user feedback, whether the processing was defective, and the determined defect cause.

Abstract: Systems and methods are described for replacing image data from a virtual machine with local image data, such as for a video application. In some aspects, a method for replacing image data in a video application may include obtaining first image data from a camera connected to the user device and transmitting the first image data to a virtual machine that executes the video application. The user device may receive a video application image from the virtual machine that includes an indication that identifies the first image data. The user device may identify the first image data in the video application image based on the indication and replace it with second image data obtained from the camera device to produce a modified video application image. The client device may render the modified video application image in a virtual computing environment displayed on the user device.

Abstract: A physical space contains stationary objects that do not move over time (e.g., a couch) and may have non-stationary objects that do move over time (e.g., people and pets). An autonomous mobile device (AMD) determines and uses an occupancy map of stationary objects to find a route from one point to another in a physical space. Non-stationary objects are detected and prevented from being incorrectly added to the occupancy map. Point cloud data is processed to determine first candidate objects. Image data is processed to determine second candidate objects. These candidate objects are associated with each other and their characteristics assessed to determine if the candidate objects are stationary or non-stationary. The occupancy map is updated with stationary obstacles. During navigation, the occupancy map may be used for route planning while the non-stationary objects are used for local avoidance.

Abstract: Techniques are disclosed for detecting an uncovered portion of a body of a person in a frame of video content. In an example, a first machine learning model of a computing system may output a first score for the frame based on a map that identifies a region of the frame associated with an uncovered body part type. Depending on a value of the first score, a second machine learning model that includes a neural network architecture may further analyze the frame to output a second score. The first score and second score may be merged to produce a third score for the frame. A plurality of scores may be determined, respectively, for frames of the video content, and a maximum score may be selected. The video content may be selected for presentation on a display for further evaluation based on the maximum score.

Abstract: Techniques for predicting a pair of an enzyme primary sequence and a substrate, and interaction probability for the pair are described. An exemplary method includes receiving a request to predict a pair of an enzyme primary sequence and a substrate, and interaction probability for the pair; combining an enzyme vector, a substrate vector, and an interaction indication for the enzyme and substrate to form a machine learning model input; applying a machine learning model to the machine learning model input to predict the pair of an enzyme primary sequence and a substrate, and interaction probability for the pair; and outputting a result of the application of the machine learning model including the predicted pair of an enzyme primary sequence and a substrate, and interaction probability for the pair.

Abstract: Techniques for communications in a multi-zone space are described. In an example, a system stores data indicating devices in zones of a space and data indicating zones that are associated with a communications session. A display in a zone presents graphical user interface indicating the zones associated with the communications session. Incoming audio data can be mixed with audio data detected from another participating zone and a speaker in the zone can output this audio data. Audio data detected from the participating zone can also be mixed and sent to a remote device that is an endpoint of the communications session.

Abstract: Systems and methods are described for streaming data between a user device and a remote computing environment via a “switchboard” service that enables interaction without the user device or the remote computing environment establishing additional connections. A first routing device receives a connection from a user device that requests routing a data stream to or from a remote computing environment. The first routing device processes the request by generating a token, which is passed to the remote computing environment along with the request. The remote computing environment passes the token to a second routing device, which decodes the routing token to identify the first routing device. The second routing device then passes the request token to the first routing device, which responds by establishing a route for streaming data between the connection made with the user device and the remote computing environment via the routing devices.

Abstract: Techniques are described for providing an access control list (ACL) assisted process for filtering the analysis and display of dependency relationships among software application components (e.g., packages, files, classes, etc.), e.g., as part of modernization processes aimed at decomposing monolithic applications, identifying anti-patterns, or otherwise analyzing such applications. A software modernization service of a cloud provider network provides discovery agents and other tools that are capable of creating an inventory of users' software applications and collecting application artifacts (e.g., source code or bytecode files) associated with the software applications in users' computing environments. Various techniques are described for using ACLs containing entries enabling or disabling the analysis or display of various application components to customize various modernization processes and results displays.

Abstract: An IoT service of a provider network may be used to optimize downlink scheduling for edge devices that use a radio communication protocol (e.g., LoRaWAN). A processing instance receives DL packets from a client application that are to be sent to an edge device. The processing instance adds metadata entries (device ID, message ID) to a metadata queue and stores the DL packets into a data store. A scheduling instance obtains a metadata entry and the associated DL packet, formats the DL packet based on edge device information, and publishes the packet to a gateway topic. A separate connection instance receives the formatted DL packet and sends it to a gateway of the client network.

Abstract: Techniques for predicting a protein sequence are described. An exemplary method includes receiving a request to predict a missing area of a protein's primary sequence and a corresponding three-dimensional position of the missing area; applying a machine learning model to backbone Cartesian coordinates of the protein's primary sequence and a protein vector of a representation of the protein's primary sequence including the missing area to predict a missing area of the protein primary sequence and a corresponding three-dimensional position for the missing area, wherein the machine learning model is selected from the group consisting of: an attention-based machine learning model, a bidirectional long short term memory-based model, and a convolutional neural network-based model; and outputting a result of the machine learning model.

Abstract: A semiconductor package can include a unitary substrate defining a first region and a second region. A first interposer can be mounted on the unitary substrate in the first region, and a first die can be mounted on the first interposer. A second interposer can be mounted in the second region on the unitary substrate, and a second die can be mounted on the second interposer.

Abstract: Techniques for managing connections in a mesh network that includes multiple provisioners are described. In an example, a computer system receives, from a first device set up as a first provisioner in the mesh network, a first identifier of a third device of the mesh network. The computer system also receives, from a second device set up as a second provisioner in the mesh network, the first identifier of the third device. The computer system generates an association between the third device and the first device based at least in part on a determination that the third device is to be reached exclusively via the first device. The computer system also generates a message about an action to be performed by the third device and sends, based at least in part on the association, the message to the first device.

Abstract: A system may use multiple speech interface devices to interact with a user by speech. All or a portion of the speech interface devices may detect a user utterance and may initiate speech processing to determine a meaning or intent of the utterance. Within the speech processing, arbitration is employed to select one of the multiple speech interface devices to respond to the user utterance. Arbitration may be based in part on metadata that directly or indirectly indicates the proximity of the user to the devices, and the device that is deemed to be nearest the user may be selected to respond to the user utterance.

Abstract: A pair of virtualized security device initialization data sets are received at a first virtualization server from respective sources prior to a launch of a compute instance at the server. A first virtualized security device is initialized using the data sets, and used to generate cryptographic artifacts used by the compute instance. A data item which was included in one of the data sets is modified after the cryptographic artifacts are generated. Additional cryptographic artifacts are generated by a second virtualized security device at a second virtualization server to which the compute instance has been migrated. The second virtualized security device is initialized using at least the modified data item.

Abstract: When a client requests a data import job, a remote storage service provider provisions a shippable storage device that will be used to transfer client data from the client to the service provider for import. The service provider generates security information for the data import job, provisions the shippable storage device with the security information, and sends the shippable storage device to the client. The service provider also sends client-keys to the client, separate from the shippable storage device (e.g., via a network). The client receives the device, encrypts the client data and keys, transfers the encrypted data and keys onto the device, and ships it back to the service provider. The remote storage service provider authenticates the storage device, decrypts client-generated keys using the client-keys stored at the storage service provider, decrypts the data using the decrypted client-side generated keys, and imports the decrypted data.

Abstract: This disclosure is directed to using cascading algorithms to automatically identify items placed in a tote or other receptacle utilized by users in material handling facilities as the users move around the facilities. A tote may store a database or “gallery” of item representations for all of the items that are stored in the facility that a user may place in their totes. The tote may use multiple algorithms in a cascading manner to analyze the gallery of item representations in order to iteratively narrow the search space of item representations in the gallery to determine which of the items was placed in the tote by a user. Upon identifying the item placed in the tote, the tote may add an item identifier for the item to a virtual listing of item identifiers representing items previously placed in the tote.

Abstract: Lease-based consistency may be implemented for databases to handle failovers. A database node may obtain a consistency lease that describes a point in time determined from a time-to-live amount added to a consistent point in time for database data. While the consistency lease is valid, Multi-version Concurrency Control (MVCC) snapshots assigned by the database node can be used to handle requests to access the database data. Once expired, the database node may have to renew the consistency lease in order to continue to handle write and read requests.

Abstract: Techniques are generally described for fusing sensor data of different modalities using a transformer. In various examples, first sensor data may be received from a first sensor and second sensor data may be received from a second sensor. A first feature representation of the first sensor data may be generated using a first machine learning model and a second feature representation of the second sensor data may be generated using a second machine learning model. In some examples, a modified first feature representation of the first sensor data may be generated based at least in part on a self-attention mechanism of a transformer encoder. The modified first feature representation may be generated based at least in part on the first feature representation and the second feature representation. A computer vision task may be performed using the modified first feature representation.

Abstract: Disclosed are systems, methods, and apparatus of an automated and self-service kiosk that allows customers to select inventory items available from the kiosk and walk or move away with selected inventory item(s) without having to process payment, identify the inventory item(s), or provide any other form of checkout. After a customer has picked one or more items and departed the kiosk, the picked items are determined and the customer charged for the items. For example, one or more of detected weight changes measured at the kiosk and/or images generated at the kiosk may be used to identify items picked by the customer from the kiosk.

Abstract: A multi-assistant speech-processing system that centrally determines multiple execution plans to respond to a user input. A central component determines whether a particular input should be processed using a requested assistant or a different assistant or should be terminated. Assistant handoff may be determined based on system policies as well as user input-specific data. A ranked list of execution options may be supplemented by augmented data corresponding to messages to a user. The system may attempt to execute plans in the ranked order until a plan succeeds.

Abstract: A satellite antenna ground station service includes a plurality of ground stations and associated data centers, wherein the data centers are part of a provider network. Clients may reserve satellite antenna access time-slots via a user interface of the satellite antenna ground station service and store data directly to a data center of the provider network or to the client's premises via a direct connection between the client and the provider network. In some embodiments, the provider network may offer a plurality of network-based services, such as a compute service, a data storage service, a machine learning service, or a data analytics service, and a client may utilize one or more of these services to analyze and process downlinked data received from a satellite of the client via a satellite antenna ground station of the satellite antenna ground station service of the provider network.

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 a corresponding action, such as 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. In some instances, the speech recognition platform engages in a back-and-forth dialog with the user in order to properly fulfill the user's request.

Abstract: A safety rated robotic motor control system may be configured to receive sensor data from sensors positioned on a motor. The data can be received by a programmable logic device and processed to generate a fault message based on a first set of characteristics. The fault message and the data can be transmitted to a safety controller that determines a second set of characteristics and generates operating instructions based on the fault message and the second set of characteristics.

Abstract: Where an event is determined to have occurred at a location within a vicinity of a plurality of actors, imaging data captured using cameras having the location is processed using one or more machine learning systems or techniques operating on the cameras to determine which of the actors is most likely associated with the event. For each relevant pixel of each image captured by a camera, the camera returns a set of vectors extending to pixels of body parts of actors who are most likely to have been involved with an event occurring at the relevant pixel, along with a measure of confidence in the respective vectors. A server receives the vectors from the cameras, determines which of the images depicted the event in a favorable view, based at least in part on the quality of such images, and selects one of the actors as associated with the event accordingly.

Abstract: Commercial interactions with non-discretized items such as liquids in carafes or other dispensers are detected and associated with actors using images captured by one or more digital cameras including the carafes or dispensers within their fields of view. The images are processed to detect body parts of actors and other aspects therein, and to not only determine that a commercial interaction has occurred but also identify an actor that performed the commercial interaction. Based on information or data determined from such images, movements of body parts associated with raising, lowering or rotating one or more carafes or other dispensers may be detected, and a commercial interaction involving such carafes or dispensers may be detected and associated with a specific actor accordingly.

Abstract: Devices and techniques are generally described for iterative theme discovery in text. In some examples, first text data that is separated into a plurality of documents may be received. In some examples, a codebook comprising a first topic associated with a first set of keywords and a second topic associated with second set of keywords may be identified. Instructions to modify the codebook to generate a modified codebook may be received. The instructions may be effective to add to, delete from, and/or modify at least one of the keywords of the first set of keywords or the first topic. In some further examples, a first document of the plurality of documents may be tagged with the first topic based at least in part on the first keywords of the modified codebook. In some examples, output data may be generated that indicates that the first document pertains to the first topic.