Abstract: A structured document analyzer that associates keys and values in structured documents based on key, value, and key-value container bounding boxes. A trained machine learning model analyzes images of structured documents to determine bounding boxes for keys, values, and key-value containers in the images with confidence scores for the classifications. For each image, duplicate bounding boxes are removed, and then a set of key-value containers are selected and sorted based on the confidence scores. For each key-value container, a best key and value are determined for the container based on overlap of the key and value bounding boxes with the container bounding box and the confidence scores. Optical character recognition may be performed on the image to determine text for the keys and values.

Abstract: The dynamic selection of players for a gaming session, or users for an application session, can be based at least in part upon player relationship data. Social graph data can be maintained that indicates game experience that particular users have shared in prior game sessions. This relationship data can be used with other selection criteria, such as similar skill level or complementary play characteristics, to select players for a gaming session that are familiar to a specific player, enabling the player to have recurring experience with similar players in order to build relationships with those players and help to develop a sense of community within the game, which can improve the user experience and thus utilization of the game. Other factors can be considered as well, such as whether players have connected or blocked each other through a social network or gaming environment.

Abstract: A user can utter a voice command in an environment where multiple audio playback devices are located to have audio output on a single device, or a predefined group of devices in a synchronized manner. In instances when the voice command uttered by the user does not specify a target for audio output, an implicit target selection algorithm can evaluate one or more criteria to determine an appropriate target for output of the audio corresponding to the voice command. An example criterion is met if a predetermined time period has lapsed since a last utterance was detected by a device in the environment. However, other criteria can be evaluated for determining a target output device(s).

Abstract: In a battery-powered audio/video recording and communication device having a camera, the length of a preset interval (the interval between instances of the device sending a request to a network to check whether any user requests to access the camera have been received) may be adjusted upward or downward in order to balance the competing interests of reducing latency (e.g. reducing the delay that the user may experience when trying to access the camera) and conserving battery life. The present embodiments advantageously balance these competing interests by initially setting the length of the preset interval to be relatively short to reduce latency, but automatically increasing the length of the preset interval as the battery charge is depleted, thereby extending battery life. The present embodiments also advantageously enable the length of the preset interval to be decreased after the battery is recharged, thereby reducing latency.

Abstract: Disclosed are various embodiments or a robotic gripper. A mechanical gripper is affixed to an end effector of the robotic arm. A mount is also affixed to the mechanical gripper. An air nozzle may be affixed to the mount. The air nozzle can include an egress directed towards the mechanical gripper.

Abstract: Disclosed are various embodiments for providing multi-factor authentication credentials. In one embodiment, in response to a request from an application, a notification is generated in a notification area of a display. Entry of a user approval is facilitated via the notification. The security credential may be shown in the notification area so that a user may enter it in a form field of the application.

Abstract: Generally described, one or more aspects of the present application correspond to a machine learning address normalization system. A system of deep learning networks can normalize the tokens of a free-form address into an address component hierarchy. Feature vectors representing various characters and words of the address tokens can be input into a bi-directional long short term memory network (LSTM) to generate a hidden state representation of each token, which can be individually passed through a softmax layer to generate probabilistic values of the token being each of the components in the address hierarchy. Thereafter, a conditional random field (CRF) model can select a particular address component for each token by using learned parameters to optimize a path through the collective outputs of the softmax layer for the tokens. Thus, the free-form address can be normalized to determine the values it contains for different components of a specified address hierarchy.

Abstract: A technology is described for operating a device shadowing service that manages device representations for physical devices that are network addressable. An example method may include associating a device representation with a representation type, where the device representation represents a copy of a state of a physical device in network communication with a device shadowing service that maintains the device representation, and the representation type may define a command that can be invoked through the device representation. After the device representation has been associated with the representation type, the command may be invoked through the device representation. Invoking the command may involve identifying the representation type associated with the device representation and determining that the command is supported by the representation type. Thereafter, the command may be executed.

Abstract: Systems, methods, and devices for establishing communications sessions with contacts are disclosed. In some embodiments, a first request may be received from a first device. The first request may be to communicate with a contact name. A user account associated with the first device may then be identified, and a contact list associated with the user account may be accessed to determine contacts associated with the contact name. Based on the contact list, a first contact and a second contact associated with the contact name may be identified. It may be determined, from memory, that the first contact is a first preferred contact. However, based on an intervening event, the second contact, rather than the preferred contact, may be selected for communicating with the contact.

Abstract: This disclosure describes a system, method, and apparatus for efficiently tracking inventory using RFID tags (e.g., RFID tags). For example, an RFID tag reader may perform an inventory mode to receive RFID tag identifiers from RFID tags located within a range of an antenna element of the RFID tag reader. Those RFID tag identifiers may be included on an inventory table associated with the antenna element. In a second mode, the RFID tag reader may perform a targeted read. During a targeted read mode, the RFID tag reader, utilizing the antenna element, may issue a request for a selected RFID tag identified in the inventory table to reply to the request.

Abstract: This disclosure describes a configuration of an unmanned aerial vehicle (UAV) that will facilitate extended flight duration. The UAV may have any number of lifting motors. For example, the UAV may include four lifting motors (also known as a quad-copter), eight lifting motors (octo-copter), etc. Likewise, to improve the efficiency of horizontal flight, the UAV also includes a pushing motor and propeller assembly that is oriented at approximately ninety degrees to one or more of the lifting motors. When the UAV is moving horizontally, the pushing motor may be engaged and the pushing propeller will aid in the horizontal propulsion of the UAV.

Abstract: Techniques are described for encoding noisy media content. A residual coefficient matrix representing differences in image content between a portion of a target image frame and portions of one or more reference frames can include noise within a high frequency band. Some of the noise can be removed by removing isolated residual coefficients. Some of the noise can be reduced by attenuating their values selectively.

Abstract: In a memory controller, a prefetch indication can be sent to memory to prepare the memory for a potential future read or write. Statistics can be used to select when such a prefetch should occur. The prefetch can occur without any read or write command being commenced. As a result, the memory controller predicts when to perform the prefetch. Some examples of when a prefetch can be sent include when there are other requests for the same memory page, or how often the page is requested. The page can remain open to prevent it from closing until the relevant read or write arrives. In the case that a read or write does not occur after a predetermined period of time, then a precharge can be performed to release the memory page.

Abstract: Poses or gestures of actors within a scene may be detected and tracked using multiple imaging devices aligned with fields of view that overlap at least in part. Images captured by the imaging devices may be synchronized and provided to a classifier to recognize body parts within the images, and score maps indicative of locations of peak probabilities that the images include the respective body parts may be generated. Locations of peak values within the score maps may be correlated with one another to confirm that a given body part is depicted in two or more fields of view, and vectors indicative of distances to or ranges of motion of body parts, with respect to the given body part, may be generated. Motion of the body parts may be tracked in subsequent images, and a virtual model of the body parts may be generated and updated based on the motion.

Abstract: Finger assemblies at the end of a robotic arm end effector includes talon that are retractable. Retraction may be accomplished by a vertical or horizontal barrel cam, a paddle and spin assembly, and/or a resilient plate. A method includes contacting a portion of a finger body against a datum.

Abstract: Systems and methods to rapidly create, view, and modify three-dimensional experiences may include a two-dimensional content editing device and application and a three-dimensional experience viewing device and application. Using the two-dimensional content editing application, two-dimensional objects may be created, and properties of the two-dimensional objects may be defined. Using the three-dimensional experience viewing application, the two-dimensional objects may be rendered within a three-dimensional environment based on the defined properties. In this manner, three-dimensional experiences may be quickly created, viewed, modified, reviewed, and published without the need for specialized tools, training, or experience in three-dimensional modeling or programming.

Abstract: This disclosure describes, in part, techniques for identifying user-specific content items and/or time-based content items, as well as techniques for providing the content items to the user. For instance, a remote system may receive first audio data from an electronic device, where the first audio data represents a request to identify content related to a topic. The remote system can then store the topic in a database and use the topic to identify the content. Later, the remote system can then receive second audio data from the electronic device, where the second audio data represents a request for the content. The remote system can then send third audio data to the electronic device that represents at least a portion of the content.

Abstract: In one embodiment, a shelving unit has first and second sides, a frame, and a barrier. The frame defines a storage bay between the first and second sides, a first opening at the first side, and a second opening at the second side. The second opening receives a storage bin therethrough and into the storage bay. The first opening receives inventory items therethrough and into the storage bin when the storage bin is disposed in the storage bay. The barrier moves between (i) an open position, where the first opening is open to the storage bay so as to allow inventory items to be passed therethrough and into the storage bin, and (ii) a closed position, where the barrier at least partially obstructs at least one of the first and second openings so as to impede a human from reaching through the storage bay when the storage bin is removed.

Abstract: Some aspects of the present embodiments provide an A/V recording and communication device that stores the identification data for one or more known persons and/or objects at a local memory of the device. Each time the A/V recording and communication device detects a person or an object, the A/V recording and communication device may capture identification data for the person or object and identify the detected person or object by comparing the identification data for the detected person or object (e.g., an image of the person and/or other biometric data related to the person) with the previously stored identification data.

Abstract: A task definition is received. The task definition indicates at least a location from which one or more software image can be obtained and information usable to determine an amount of resources to allocate to one or more software containers for the one or more software image. A set of virtual machine instances in which to launch the one or more software containers is determined, the one or more software image is obtained from the location included in the task definition and is launched as the one or more of software containers within the set of virtual machine instances.

Abstract: A method and system to provide fonts for electronic publications is described. An example method includes receiving a request for an electronic publication from a user device, retrieving a font identifier of a font from the electronic publication, providing the electronic publication to the user device, determining, based on the font identifier, that at least one font file of the font is not stored on the user device, and providing the at least one font file of the font to the user device.

Abstract: A system configured to determine interactive elements for non-traditional computing devices is disclosed. The system receives a portion of a webpage with a first interactive element and an element identifier. The system determines that a threshold number of interactive actions are associated with a second interactive element in a data store. The threshold number of interactive actions indicates that the second interactive element is selectable to perform an action in response to an interaction by a user. The system determines that the first interactive element is similar to the second interactive element by a similarity measure. The system provides to a computer, a selectable option associated with the element identifier of the first interactive element. The selectable option is for selection by an indirect input to the computer and for causing the action associated with the first interactive element upon the selection of the selectable option.

Abstract: Consistent distributed edge models via controlled dropout model training are described. According to some embodiments, a machine learning model is trained using multiple sensor data streams. During the training process, ones of the sensor data streams are dropped to cause the model to be generated to be robust in that it can tolerate missing input data from sensor data sources yet still maintain high accuracy. The model can be deployed to multiple sensor devices within an environment. The sensor devices generate sensor data and exchange a variety of types of data to ultimately result in a distributed, consistent model result being generated that remains accurate despite communication faults that may occur between ones of the sensor devices.

Abstract: Interactive compilation of software to a hardware language may be performed to satisfy formal verification constraints. Source code for software to be executed on a hardware design may be received. Intermediate code may be generated from the source code as part of translating the source code to a hardware language used to specify the hardware design. The intermediate code may be provided via an interface and updates to the intermediate code may be received. The updated source code may then be used to complete translation of the source code to the hardware language.

Abstract: Techniques and systems for storing and retrieving data storage devices of a data storage system are disclosed. In some embodiments, inventory holders are used to store data storage devices used by a data storage system. When data is to be transacted with the data storage devices, mobile drive units locate appropriate inventory holders and transport them to a device reading station, where an appropriate device retrieval unit transacts the data. After the data has been transacted, the data storage devices are returned to the appropriate inventory holders, and the inventory holders are placed by the mobile drive units in locations where they may be accessed in response to further data transactions.

Abstract: A service provider receives a request to obtain a text-based transmission. In response to the request, the service provider generates an identifier corresponding to the request and, based at least in part on the identifier and text to be included in the transmission, determines a set of textual transformations to associate with the identifier. The service provider modifies the text in accordance with the set of textual transformations and provides the modified text to the requestor as the text-based transmission.

Abstract: Embodiments are disclosed for launching a virtual machine and then configuring the virtual machine after launch. Metadata configuration information can be stored in conjunction with virtual machine image configuration data. The virtual machine image configuration data can be used to launch a virtual machine. The metadata configuration information can then be used to further configure the virtual machine after launch. Both the metadata configuration information and the virtual machine image configuration data can be stored in the same virtual machine registration record.

Abstract: Techniques describing methods and systems for validating different versions of a set of rules for performing automated workflow processes. Sets of input data, each disjoint from one another, are obtained that are optimized from inputs contained in data storage. A first output is produced using one or more of the sets of inputs as input to a first version of the set of rules, and a second output is produced using the one or more of the sets of inputs as input to a second version of the set of rules. The first output is compared with the second output and, in response to a determination that the first output and the second output do not match, an indication of inconsistency is recorded.

Abstract: A system for conveying air from one location to another includes a soft duct having a passage and an air flow control device. The air flow control device can be operated to vary a cross sectional area of a portion of the passage of the soft duct.

Abstract: Technologies are provided for generation of points of insertion of directed content into a video asset. In some embodiments, multiple time offsets within an interval spanned by the video asset can be determined using audio data corresponding to the video asset. A time offset defines a boundary between first and second segments of the video asset. Using image data corresponding to the video asset, respective pairs of video clips for the multiple time offsets can be generated. Visual features, aural features, and language features pertaining to the respective pairs of video clips can then be generated. Scores for the multiple time offsets can be generated using the visual features, the aural features, and the language features. A score represents an assessment of suitability to insert directed content into the video asset at a time offset. A file that contains specific time offsets can be generated.

Abstract: This application describes techniques for determining a constituency and a layout of a group of devices within an environment. By determining the devices that form the group, as well as the positions of these devices relative to one another, the group of devices may be used to more intelligently output content within the environment.

Abstract: Bandwidth-efficient (i.e., compressed) representations of audio data can be utilized for near real-time presentation of the audio on one or more receiving devices. Persons identified as having speech represented in the audio data can have trained speech models provided to the devices. These trained models can be used to classify the compressed audio in order to improve the quality to correspond more closely to the uncompressed version, without experiencing lag that might otherwise be associated with transmission of the uncompressed audio. The uncompressed audio is also received, with potential lag, and is used to further train the speech models in near real time. The ability to utilize the uncompressed audio as it is received prevents a need to store or further transmit the audio data for offline processing, and enables the further trained model to be used during the communication session.

Abstract: A convolutional neural network may be trained to inspect subjects such as carbon fiber propellers for surface flaws or other damage. The convolutional neural network may be trained using images of damaged and undamaged subjects. The damaged subjects may be damaged authentically during operation or artificially by manual or automated means. Additionally, images of undamaged subjects may be synthetically altered to depict damages, and such images may be used to train the convolutional neural network. Images of damaged and undamaged subjects may be captured for training or inspection purposes by an imaging system having cameras aligned substantially perpendicular to subjects and planar light sources aligned to project light upon the subjects in a manner that minimizes shadows and specular reflections. Once the classifier is trained, patches of an image of a subject may be provided to the classifier, which may predict whether such patches depict damage to the subject.

Abstract: Respective statistical distributions of a target variable within a proposed training data set and a proposed test data set for a machine learning model are obtained. A metric indicative of the difference between the two statistical distributions is computed. The difference metric is used to determine whether the proposed test data set is acceptable to evaluate the machine learning model.

Abstract: Method and apparatus for stress management in a searchable data service. The searchable data service may provide a searchable index to a backend data store, and an interface to build and query the searchable index, that enables client applications to search for and retrieve locators for stored entities in the backend data store. Embodiments of the searchable data service may implement a distributed stress management mechanism that may provide functionality including, but not limited to, the automated monitoring of critical resources, analysis of resource usage, and decisions on and performance of actions to keep resource usage within comfort zones. In one embodiment, in response to usage of a particular resource being detected as out of the comfort zone on a node, an action may be performed to transfer at least part of the resource usage for the local resource to another node that provides a similar resource.

Abstract: Various embodiments of a system and method for tracking service requests are described. Embodiments may include call tree generation logic configured to receive multiple request identifiers associated with a respective one of multiple service requests. Each given request identifier may include an origin identifier, a depth value, and a request stack comprising one or more interaction identifiers. The call tree generation logic may also be configured to, based on multiple request identifiers that each include an origin identifier associated with a particular root request, generating a data structure that specifies a hierarchy of services called to fulfill that particular root request. Based on one or more of the interaction identifiers and one or more of the depth values, the generated data structure may specify for each given service of the hierarchy: a parent service that called the given service, and one or more child services called by the given service.

Abstract: Techniques are described for providing users with access to computer networks, such as to enable users to interact with a remote configurable network service in order to create and configure computer networks that are provided by the configurable network service for use by the users. Computer networks provided by the configurable network service may be configured to be private computer networks that are accessible only by the users who create them, and may each be created and configured by a client of the configurable network service to be an extension to an existing computer network of the client, such as a private computer network extension to an existing private computer network of the client. If so, secure private access between an existing computer network and new computer network extension that is being provided may be enabled using one or more VPN connections or other private access mechanisms.

Abstract: A technology is described for distributing a publication message. An example method may include receiving a message at a first device gateway to broadcast to devices subscribed to a broadcast topic. The first device gateway maintains a first subscription table indicating which devices connected to the first device gateway are subscribed to the broadcast topic. The first device gateway distributes the message to the devices indicated in the first subscription table as being subscribed to the broadcast topic. In addition, the first device gateway may distribute the message to a second device gateway which maintains a second subscription table indicating which devices connected to the second device gateway are subscribed to the broadcast topic, and the second device gateway distributes the message to the devices indicated in the second subscription table as being subscribed to the broadcast topic.

Abstract: At a network manager of an extension resource group of a provider network, a message comprising a command to launch a compute instance is received at an address which is part of a first network configured at a premise external to the provider network. The extension resource group includes a first host at the external premise. Within a second network configured at the external premise, the first host is assigned an address within a second address range. Addresses within the second range are also assigned to hosts within the provider network. The command is transmitted to the first host, and a compute instance is instantiated.

Abstract: A database may provide a visualization of restoration operation granularity. Information may be obtained describing possible restoration points within a given period of time for a database. The information may include a respective state of the database associated with a point in time and a respective restoration operation type for the possible restoration points. A graphical representation may be generated illustrating the possible restoration points ordered within the given time period according to the points in time associated with the respective states of the database. The graphical representation may be provided via network-based interface to a client of the database. The graphical representation of the possible restoration points may be updated in response to changes to the given time period or database volume events that modify the possible restoration points.

Abstract: A method and system for generating access policy suggestions are described. The system initiates a test mode for a web service to be access by a user. The system authorizes access to the web service by the user in the test mode and records test data regarding the access to the web service by the user in the test mode. The system generates an access policy suggestion based on the test data.

Abstract: Techniques are described for associating identifiers (e.g., digital watermarks) with video content in a way that enables identification of the source of pirated content with specificity as granular as an individual user account. A compositors operating in the DRM trust zone of a client device introduces the identifier by compositing overlay information with decoded video frames. The identifier may then be recovered by comparing target content to the source content to extract the overlay information.

Abstract: Features are disclosed for detecting words in audio using contextual information in addition to automatic speech recognition results. A detection model can be generated and used to determine whether a particular word, such as a keyword or “wake word,” has been uttered. The detection model can operate on features derived from an audio signal, contextual information associated with generation of the audio signal, and the like. In some embodiments, the detection model can be customized for particular users or groups of users based usage patterns associated with the users.

Abstract: A switching assembly selectively routes power received from upstream power distribution systems over power transmission lines to a downstream electrical load using data signals communicated to the assembly from the power distribution systems over the power transmission lines. The switching assembly monitors information in the data signals to determine conditions in the upstream power distribution systems. The switching assembly can use the conditions in the power distribution systems and monitored power feed characteristics of power received at the switching assembly to determine whether to command a switching mechanism to switch the power feed routed to the load. Power system conditions can be used to adjust tolerance thresholds against which power feed characteristics are compared to determine whether to command a switch. The switching assembly can include an interposer device that controls the switching mechanism based on analysis of data signals received over the power transmission line.

Abstract: Techniques for packaging and deploying algorithms utilizing containers for flexible machine learning are described. In some embodiments, users can create or utilize simple containers adhering to a specification of a machine learning service in a provider network, where the containers include code for how a machine learning model is to be trained and/or executed. The machine learning service can automatically train a model and/or host a model using the containers. The containers can use a wide variety of algorithms and use a variety of types of languages, libraries, data types, etc. Users can thus implement machine learning training and/or hosting with extremely minimal knowledge of how the overall training and/or hosting is actually performed.

Abstract: Disclosed are various embodiments for malware detection by way of proxy servers. In one embodiment, a proxied request for a network resource from a network site is received from a client device by a proxy server application. The proxied request is analyzed to determine whether the proxied request includes protected information transmitted in an unsecured manner. It is then determined whether the network resource comprises malware based at least in part on an execution of the network resource or whether the proxied request includes the protected information transmitted in the unsecured manner. The proxy server application refrains from sending data generated by the network resource to the client device in response to the proxied request when the network resource is determined to comprise the malware.