Abstract: Different types of index structures are used for a replica group of a database. A leader node of a replica group performs receives updates to a copy of the database using a first type of index structure. A follower node performs updates received from the leader node as a log of updates to a copy of the database in an external storage system when a size of the received updates exceeds a threshold. The follower node performs requests to read data from the database using the copy in the external storage.

Abstract: An optical communication system comprises a plurality of linked single-wavelength coherent optical transceivers configured to communicate via short-reach data-center links at a common reference wavelength, each transceiver including a single single-wavelength laser source that is used to produce a source beam at a source beam wavelength, the source beam being used to produce transmit beams sent by the transceiver and for mixing with one or more receive beams received by the transceiver; wherein each of the transceivers includes a thermo-electric cooler configured to control a temperature of the single-wavelength laser source around a temperature setpoint that locks the source beam wavelength to the common reference wavelength; wherein each transceiver includes a wavelength deviation detector configured to detect a wavelength deviation between the source beam wavelength and the wavelength or wavelengths of the one or more receive beams or between the source beam wavelength and the common reference wavelength,

Abstract: Techniques for calculating and using a request cost index for throttling application programming interface (API) requests to execute operations in a provider network. The techniques encompass the step receiving a request to execute an operation at an API service in the provider network. Further steps include determining to execute the operation based on a request rate limiting algorithm, executing the operation to yield an operation result, and sending the operation result. Additional steps include calculating a request cost index that reflects an amount of computing resources utilized by executing the operation, determining an adjustment amount for a state variable of the request rate limiting algorithm based on the calculated request cost index, and adjusting (e.g., lowering) the state variable by the adjustment amount. Other (e.g., subsequent) requests to execute queries received at the API service that are metered by the state variable can be throttled by the API service.

Abstract: Computer systems and associated methods are disclosed to implement a model development environment (MDE) that allows a team of users to perform iterative model experiments to develop machine learning (ML) media models. In embodiments, the MDE implements a media data management interface that allows users to annotate and manage training data for models. In embodiments, the MDE implements a model experimentation interface that allows users to configure and run model experiments, which include a training run and a test run of a model. In embodiments, the MDE implements a model diagnosis interface that displays the model's performance metrics and allows users to visually inspect media samples that were used during the model experiment to determine corrective actions to improve model performance for later iterations of experiments. In embodiments, the MDE allows different types of users to collaborate on a series of model experiments to build an optimal media model.

Abstract: A computing system may detect that a service, which receives a plurality of task requests associated with clients or profiles, is under duress based on performance information associated with the service. The computing system, responsive to detecting that the service is under duress, may successively, until the service is detected to not be under duress, select a profile based on a respective volume of requests associated with the profile, apply a task request limit to the profile, and detect whether the service is still under duress.

Abstract: A system configured to perform content identification using fingerprinting to recognize known media content. The system may generate a reference database including reference fingerprints for each media content item to include in the content identification. In addition, the system may generate a hash table that associates individual frames of the reference fingerprints with identification information for corresponding media content items. When a device is playing media content, the system may perform content identification by generating query fingerprints representing the media content and comparing the query fingerprints to the reference database. For example, the system may match a query fingerprint to a reference fingerprint by identifying which of the reference fingerprints shares the most frames with the query fingerprint using the hash table. In addition, the system may use additional decision criteria to confirm a match, such as fine-grain matching or tracking successive fingerprints over time.

Abstract: Portions of text data generated from inverse text normalization may be redacted. Text data for redaction may be obtained. One or more inverse text normalization models may be applied to the text data to generate normalized text data. A machine learning model, trained to recognize text for redaction, may be applied to identify portions of the normalized text data for redaction. The identified portions may be redacted and the redacted normalized text provided to a destination.

Abstract: A system, technique, or computer program product generates a simplified version of a geometry, based on a target number of points to be included in the output. A first plurality of points, representative of a geometry, is received. The simplified version of the geometry is generated by, at least, expanding a segment of a simplified version of the geometry. The segment is identified for expanding by determining that a point associated with the segment is at a distance from the segment that exceeds a tolerance value, and is includable in the simplified version of the geometry without causing the simplified version of the geometry to exceed the target size.

Abstract: It is determined that a workflow comprising inter-resource requests of a distributed application is to be initiated. In response to detecting, at a first resource during execution of the workflow, that a triggering condition for retrying a request is met, and that a workflow-level retry budget of the workflow indicates that a retry is permitted, the retry is attempted and the budget is modified to indicate that the retry has been attempted. In response to detecting, at another resource, that a triggering condition for retrying another request is met, and that the workflow-level retry budget of the workflow indicates that a retry is not permitted, an indication that the workflow has failed is generated.

Abstract: Systems and methods are provided for creating and running an instance of a dynamic access control system (DACS). Trust providers may be defined in a trust broker of the DACS such that trust information associated with the trust providers can be used to create a custom data structure. Resources and resource groups may be defined in the DACS. Policies may be configured or coded in the DACS to map the custom data structure to recourses or resources groups. Additionally, policies may be configured or coded in the DACS to route the data structure and request to network segments or shared with other parties.

Abstract: Systems, devices, and methods are provided for package delivery guidance and assistance. A vehicle may comprise one or more sensors. A processing unit of the vehicle may determine, based on vehicle data obtained from the one or more sensors, that the vehicle is in a first state, determine that the first state corresponds to a first sub-task of the plurality of sub-tasks, determine a first graphical interface for performance of the first sub-task, present the first graphical interface on a first display screen of the at vehicle, responsive to additional vehicle data obtained from the one or more sensors indicating that the vehicle is in a second state: determine that the second state corresponds to a second sub-task of the plurality of sub-tasks, and update the first display screen with a second graphical interface that is for performance of the second sub-task.

Abstract: A systolic array can implement an architecture tailored to perform matrix multiplications on sparse matrices. Each processing element in the systolic array may include a register configured to store a value, and a multiplexor configured to select an input element from multiple input data buses based on metadata associated with the value. Each processing element may also include a multiplier configured to multiply the selected input element with the value to generate a multiplication result, and an adder configured to add the multiplication result to a partial sum input to generate a partial sum output.

Abstract: A camera device may capture a high-resolution image of a frame and store the high-resolution image in memory. The camera device may down-sample the high-resolution image to a low-resolution image of the frame. The camera device may transmit, to an external compute node, the low-resolution image. The camera device may receive, from the external compute node, a request for a region of interest from the high-resolution image. The camera device may transmit, to the external compute node, the region of interest from the high-resolution image. The camera device may have a captured frame rate at which high-resolution images are captured. The camera device may also have an external frame rate based upon which all, or only some, of the high-resolution images may be down-sampled to low-resolution images that are transmitted to the external compute node. The external frame rate may be decoupled from the captured frame rate.

Abstract: Systems and methods are provided to enable parallelized multiply-accumulate operations in a systolic array. Each column of the systolic array can include multiple busses enabling independent transmission of input partial sums along the respective bus. Each processing element of a given columnar bus can receive an input partial sum from a prior element of the given columnar bus, and perform arithmetic operations on the input partial sum. Each processing element can generate an output partial sum based on the arithmetic operations, provide the output partial sum to a next processing element of the given columnar bus, without the output partial sum being processed by a processing element of the column located between the two processing elements that uses a different columnar bus. Use of columnar busses can enable parallelization to increase speed or enable increased latency at individual processing elements.

Abstract: Techniques for utilizing machine learning to generate and use publishing parameters for a content delivery system are described.

Abstract: Methods and apparatuses for hierarchical partitioning of operators of a neural network for execution on an acceleration engine are provided. Neural networks are built in machine learning frameworks using neural network operators. The neural network operators are compiled into executable code for the acceleration engine. Development of new framework-level operators can exceed the capability to map the newly developed framework-level operators onto the acceleration engine. To enable neural networks to be executed on an acceleration engine, hierarchical partitioning can be used to partition the operators of the neural network. The hierarchical partitioning can identify operators that are supported by a compiler for execution on the acceleration engine, operators to be compiled for execution on a host processor, and operators to be executed on the machine learning framework.

Abstract: This disclosure describes, in part, techniques for adapting bandwidth estimation algorithms to account for local network stalls (e.g., Wi-Fi stalls). For instance, an electronic device may use a pacer component and/or a network socket buffer to detect Wi-Fi stalls. The pacer component and/or the network socket buffer may both be configured to store packets. The pacer component may determine a number of bytes to send to the network socket buffer based on a current bandwidth (or bitrate) estimation value and may move packets from the pacer component to the network socket buffer. If the second queue reaches capacity at a first time and is no longer at capacity at a second subsequent time, a Wi-Fi stall may be detected, and the electronic device may transition from a first state to a second state causing the electronic device to determine whether to update a current bandwidth estimation value.

Abstract: Machine learning models may be generated that are compatible with another machine learning model and satisfy a resource constraint. Techniques that ensure weight compatibility and architectural compatibility between a machine learning model being created to be compatible with another machine learning model are applied. The resource constraint is enforced so that the generated machine learning model also fits within the resource constraint.

Abstract: Described are systems and methods for providing fault tolerant drivers for electric motors. Embodiments of the present disclosure may provide a single power electronics unit and/or ESC that is configured to drive more than one electric motor. According to exemplary embodiments, each power electronics unit and/or ESC may be connected to more than one electric motor (e.g., three electric motors, etc.) and each power electronics unit and/or ESC may be configured to drive and/or control a single phase of multiple electric motors. This can facilitate two-phase mode operation in the event of a faulty power electronics unit and/or ESC, thereby facilitating continued operation of the affected electric motors. Exemplary embodiments of the present disclosure can also provide alternate modes of operation for the electric motors that are operating in two-phase mode to mitigate thermal stresses that may be experienced by the power electronics and/or the electric motors during two-phase operation.

Abstract: A medical audio summarization service receives a medical conversation and an indication of a user preferred summarization style selected from a plurality of available summarization styles to generate a medical summary that conforms to the user preferred summarization style. A transcript is generated via a medical audio transcription service, and the transcript is used by a natural language processing engine (including a large language model) to generate the medical summary. The large language model is trained to be used to generate medical summaries that conform to respective ones of a plurality of user preferred summarization styles. The large language model is trained using training data comprising previously generated summaries and summary interaction metadata generated from user edits and/or feedback.

Abstract: A placement plan for training state checkpoints of a machine learning model is generated based at least in part on a number of training servers of a distributed training environment. The plan indicates, with respect to an individual server, one or more other servers at which replicas of training state checkpoints of the individual server are to be stored. During selected periods of one or more training iterations of the model, respective portions of a replica of a training state checkpoint of a first server are transmitted to a second server selected based on the placement plan. After an event causes disruption of the training iterations, one of the checkpoints generated at the first server is retrieved from the second server and used to resume the training iterations.

Abstract: Systems, methods, and apparatuses for detecting anomalies using clusters are described. In some examples, a method includes receiving a request to perform anomaly detection using a plurality of clusters; receiving a data point; determining when the received data point is a part of one of the plurality of clusters utilizing a distance to centers of the one or more clusters, wherein: when the received data point is determined to belong to a normal cluster, assigning the received data point to the determined cluster, updating the cluster, and updating a history for the cluster, when the received data point is determined to belong to an anomalous cluster, raising an anomaly, updating the cluster, and updating a history for the cluster, and when the received data point is determined to not belong to any cluster, raising an anomaly.

Abstract: Techniques for providing network applications are described. For instance, system(s) may install a network application onto a virtual server. While installing the network application, the system(s) may monitor the installation in order to identify events. The system(s) may then generate a first file that includes file events, a second file that includes registry events, and a third file that includes service events. Additionally, the system may copy the software files installed on the virtual server. The system(s) may then generate a software package that includes the files and store the software package on a virtual storage device. After storing the software package, the system(s) may make copies of the software package and store the copies on multiple virtual storage devices. The system(s) may then use the virtual storage devices to install and launch the network application on virtual servers.

Abstract: Techniques are described for enabling users of a cloud provider network to create policies used to control the use of temporary security credentials by computing resources other than a computing resource to which the credentials were issued. An identity and access management service encodes, into temporary security credentials, information about the virtual private network to which the credentials are issued. When a computing resource subsequently issues requests to perform actions and uses the temporary security credentials to sign the request, the cloud provider network further adds, to the network traffic, information associated with the virtual private network from which the request originates. A user can then create a policy with a statement indicating that request are to be permitted only if, e.g., the identity of the virtual private network as encoded in the temporary security credentials matches the identity of the virtual private network identified by the information included in the request.

Abstract: Techniques are disclosed for automatically generating an access envelope that can be used to provide recommended access permissions and access rules for access roles. An access policy automation service may monitor actions requested by a user or system operating within a workflow from computing resources in a computing environment. The service may automatically update an access envelope associated with the user or system with permissions allowing the user or system to perform the requested actions. The envelope can then be used in a role definition to provide appropriate permission to users who may be performing similar operations.

Abstract: A system comprising one or more computing devices implements a vehicle information extraction service. The vehicle information extraction service enables customers to maintain a model of a fleet of vehicles and collect vehicle information from heterogenous vehicles included in the fleet, wherein the vehicles are configured with different in-vehicle communication configurations. The vehicle information extraction service automatically generates scheme packets for data collection to be sent to the heterogenous vehicles, wherein are formatted in accordance with the different respective in-vehicle communication configurations of the heterogenous vehicles, without the customer of the vehicle information extraction service having to account for these differences.

Abstract: Systems, apparatus, and techniques for automated manipulation of an adjustable container are described. An example apparatus includes a gantry and a plurality of motors. The plurality of motors are attached to the gantry. The plurality of motors are adapted to automatically adjust one or more dividers within a container to form a volume around one or more items within the container.

Abstract: A fuzzy scoreboard can compute, using a signature function, a first signature of an expected data stream associated with an input data stream that is being inputted to a design-under-test (DUT) for a datapath test. The first signature of the expected data stream can be stored without storing the expected data stream. The fuzzy scoreboard can also compute, using the same signature function, a second signature of an output data stream that is outputted from the DUT during the datapath test. The first signature can be compared with the second signature to determine whether there is a match. Storing the first signature of the expected data stream without storing the expected data stream can reduce the memory space consumed by the fuzzy scoreboard.

Abstract: Techniques for selecting a skill to execute in response to a natural language input are described. A system may receive a natural language input, determine profile data associated with the natural language input, and determine the profile data indicates a locale and at least first language and second languages. The system determines first and second sets of skills corresponding to the locale/first language and locale/second language, respectively. The system determines a first group of skill candidates corresponding to a portion of the first set of skills, and determines a second group of skill candidates corresponding to a portion of the second set of skills. The system performs ranking processing across the first and second groups of skills to determine a best skill for responding to the natural language input. Thus, in some situations, the skill invoked may not correspond to the same language represented in the natural language input.

Abstract: Systems and methods are disclosed for visual indicators for optimized packing of multi-item packages. In one embodiment, an example system may include an optionally bottomless container, a projector, and a controller configured to determine a first size of a first item to be placed in the bottomless container, determine a second size of a second item to be placed in the bottomless container, determine an item arrangement for the first item and the second item, and determine a first position for a first visual indicator. The controller may be configured to determine a second position for a second visual indicator, determine a size for a square to be projected by the projector, and cause the projector to project the first visual indicator at the first position, the second visual indicator at the second position, and the square onto the bottomless container.

Abstract: Methods and apparatus for interfaces to manage inter-regional connectivity for direct network peerings. A system may include a connectivity coordinator, a first resource collection in a first geographical zone and a second resource collection in a second geographical zone. The coordinator implements a programmatic interface defining connectivity operations. The coordinator receives a request via the interface to establish a logically isolated network path to the second resource collection on behalf of a client that has a dedicated physical link set up to connect to the first resource collection. In response to the request, the coordinator performs one or more configuration operations to enable traffic to flow from the client's network to the second resource collection over a logically isolated network path using the dedicated physical link.

Abstract: A distributed data store may maintain versioned hierarchical data structures. Different versions of a hierarchical data structure may be maintained consistent with a transaction log for the hierarchical data structure. When access requests directed to the hierarchical data structure are received, a version of the hierarchical data structure may be identified for processing an access request. For access requests with snapshot isolation, the identified version alone may be sufficient to consistently process the access request. For access requests with higher isolation requirements, such as serializable isolation, transactions based on the access request may be submitted to the transaction log so that access requests resulting in committed transactions may be allowed, whereas access requests resulting in conflicting transactions may be denied.

Abstract: Disclosed are various embodiments for provisioning radio-based networks with locality rules. In one embodiment, at least one locality rule associated with an organization is accessed. The locality rule(s) require that at least a subset of network traffic for a radio-based network remain within a particular geographic area. The radio-based network includes a radio access network and an associated core network. A topology for the radio-based network is determined based at least in part on the locality rule(s). The radio-based network is provisioned or reconfigured for the organization to have the topology complying with the at least one locality rule.

Abstract: Random cut trees are generated with respective to respective samples of a baseline set of data records of a data set for which outlier detection is to be performed. To construct a particular random cut tree, an iterative splitting technique is used, in which the attribute along which a given set of data records is split is selected based on its value range. With respect to a newly-received data record of the stream, an outlier score is determined based at least partly on a potential insertion location of a node representing the data record in a particular random cut tree, without necessarily modifying the random cut tree.

Abstract: Techniques for updating a machine learning model based on user interactions are described. In particular, in some examples, user interactions with a chatbot provide aspects of a data set to be used to train or fine-tune a ML model. In some examples, this is accomplished by collecting data from a first plurality of interactions with a machine learning (ML) model; generating a variant of the ML model using the collected data by: filtering the collected data to create a first data set, training the ML model based on the first data set to generate an adapted ML model, and fine-tuning the adapted ML model on a second data set, different than the first data set to generate the variant of the ML model.

Abstract: Virtual resources may be provisioned in a manner that is aware of, and respects, underlying implementation resource boundaries. A customer of the virtual resource provider may specify that particular virtual resources are to be implemented with implementation resources that are dedicated to the customer. Dedicating an implementation resource to a particular customer of a virtual resource provider may establish one or more information barriers between the particular customer and other customers of the virtual resource provider. Implementation resources may require transition procedures, including custom transition procedures, to enter and exit dedicated implementation resource pools. Costs corresponding to active and inactive implementation resources in a dedicated pools associated with a particular customer may be accounted for, and presented to, the customer in a variety of ways including explicit, adjusted per customer and adjusted per type of virtual resource and/or implementation resource.

Abstract: Some embodiments provide a method for training a machine-trained (MT) network. The method propagates multiple inputs through the MT network to generate an output for each of the inputs. each of the inputs is associated with an expected output, the MT network uses multiple network parameters to process the inputs, and each network parameter of a set of the network parameters is defined during training as a probability distribution across a discrete set of possible values for the network parameter. The method calculates a value of a loss function for the MT network that includes (i) a first term that measures network error based on the expected outputs compared to the generated outputs and (ii) a second term that penalizes divergence of the probability distribution for each network parameter in the set of network parameters from a predefined probability distribution for the network parameter.

Abstract: A speech-processing system may provide access to one or more virtual assistants via an audio-controlled device. A multi-agent component of a device may be used to coordinate control of device process(es) regardless of the invoked agent and without necessarily sharing information between agents and corresponding speech-processing system(s). Device-state information may be sent from the multi-agent component to a speech-processing system to interpret a user command. If the command relates to a device process control it may be routed to a device skill and to the multi-agent component of the device. The multi-agent component may track a device state and may inform a particular speech-processing system if a process related to the particular speech-processing system has been controlled.

Abstract: A coupler has a housing including a first end configured to couple to an autonomous guided vehicle (AGV) and a second end having a receptacle. A cabinet operably couples to the housing and is at least partially disposed within the receptacle. A magnetic element is disposed at least partially within the cabinet. The magnetic element is configured to magnetically engage with a portion of a towable cart for transport by the AGV.

Abstract: Methods, systems, and computer-readable media for automated selection of secure design patterns are disclosed. One or more graphs comprising a plurality of nodes and a plurality of edges are generated. At least a portion of the nodes represent software components and at least a portion of the edges represent relationships, and the one or more graphs comprise a plurality of sub-graphs. Using a graph analysis engine, the sub-graphs are analyzed for compliance with one or more security policies. A particular sub-graph is selected based at least in part on respective policy compliance of the sub-graphs. A design template is determined based at least in part on one or more software components in the particular sub-graph. The design template comprises a configuration compliant with the one or more security policies and is provided in a development environment.

Abstract: Techniques for certificate authority (CA) selection are described. A certificate management service of a cloud provider network receives a first request to generate a certificate from an electronic device, the first request including an indication of an identity of a user and an identification of a domain name to associate with the certificate. A CA selection policy applicable to the first request is identified, the CA selection policy including a CA selection rule. A CA to generate the certificate is identified by evaluating the CA selection rule, the CA selection rule associates at least a portion of the domain name with the CA. A second request to generate the certificate is sent to the identified CA. The certificate or an identification of the certificate from the CA is returned to the electronic device.

Abstract: Disclosed are various embodiments for managing security credentials for an authentication management client on a client device. In one non-limiting example, a computing device is configured to receive an authentication request from an authentication management client of a client and determine an affinity of the authentication management client based at least in part on the authentication request. The computing device is configured to determine that the authentication management client is supported based at least in part on the affinity. The computing device is configured to generate a session for the authentication management client based at least in part on a security credential being received from the authentication management client.

Abstract: Queries may be made against a non-relational database without impacting the capacities of the non-relational database by enabling performance of queries to a non-relational database via an external query management, backup management, and a non-relational database backup hosted in an external data storage. The query management and backup management of the non-relational database service may split transactional load to the non-relational database and analytical load, and furthermore allow selection of a desired data freshness for queries including various time ranges, such as seconds, minutes, or days.

Abstract: Techniques for resource-based network traffic management are described. A service of a cloud provider network receives a traffic policy, the traffic policy identifying a path to a gateway to an external network, the path identifies at least one network function in the path by a resource identifier of the cloud provider network. Traffic policy association data is received, the traffic policy association data associating the traffic policy with one or more virtual networks hosted by the cloud provider network. A network configuration of the cloud provider network is updated to route network traffic from a first virtual network to the gateway to the external network through the network function.

Abstract: This disclosure describes, in part, techniques for generating location information associated with a video. For instance, an electronic device may use a radar sensor to generate radar data. The electronic device may then analyze the radar data in order to determine a location associated with an object. In some instances, the location may correspond to at least a first coordinate along a first axis and a second coordinate along a second axis. The electronic device may also generate image data using the imaging device. Additionally, the electronic device may analyze the image data in order to determine that the image data represents an object and/or a type of object. The electronic device may then generate location data representing an identifier for the object and the location. Next, the electronic device may send the image data and the location data to one or more computing devices.

Abstract: Route tables may be associated with ingress traffic for logically isolated networks. A routing device at the edge of a logically isolated network may receive a route to include in a route table that is associated with ingress traffic to the logically isolated network, where the ingress traffic is destined for a block of public or private IP addresses. The route instructs the edge routing device to forward such ingress traffic to a network interface of a network appliance hosted in the logically isolated network. Network packets received at the edge routing device may have a destination of one or more public or private IP addresses in the block of public/private IP addresses. The edge routing device may identify the route in the route table that forwards the ingress network traffic destined for the block of public or private IP addresses to the network interface for the network appliance.

Abstract: Methods, systems, and computer-readable media for dynamic bucketization based on real-time access patterns are disclosed. Item identifiers corresponding to items in a data store are distributed among of buckets based at least in part on one or more bucket parameters. The item identifiers are divided into access windows within the buckets. Items corresponding to item identifiers in one or more access windows are loaded from the data store. Based at least in part on one or more access requests from a client with respect to the loaded items, one or more item identifiers are moved between access windows. The updated access windows with different sets of items are stored back to the data store.

Abstract: Techniques for unique label recognition and notification for streaming video. A video stream from a camera-equipped electronic device is received via the API endpoint of a stream processing service in a provider network. A request for unique label recognition and notification is received via an API endpoint of a computer vision service in the provider network. The computer vision service analyzes a video fragment of the video stream. If a unique label is recognized in the video fragment, then a notification service in the provider network sends a notification that the unique label was recognized in video captured by the camera-equipped electronic device.

Abstract: A computer-implemented method includes generating, based on a representation of a tensor mapping between an input tensor and an output tensor, a list of mappings from elements of the input tensor to elements of the output tensor, and generating groups of mappings from the list of mappings, where each of the groups of mappings corresponds to a respective set of matrix multiplications, a matrix transpose, or both. The computer-implemented method also includes generating a respective expression for each of the groups of mappings and generating code for summing results of the respective expressions, where each respective expression includes the respective set of matrix multiplications, the matrix transpose, or both.