Abstract: Various embodiments of a method and system for determining sets of variant items are described. Various embodiments may include a system configured to generate multiple item pairs each corresponding to a particular item and another item determined to be similar to the particular item. For the particular item and the other item, each item pair may include a respective sequence of text strings (e.g., a title). For each item pair, the system may perform a corresponding text alignment and determine one or more misalignments of the item pair. The system may also assign a similarity score to each item pair; the similarity score may be dependent on the misalignment(s) determined for the particular item pair. Based on each aligned item pair and the similarity score assigned to that aligned item pair, the system may generate an indication specifying that each of a set of items are variants of each other.

Abstract: A target commit sequence number (CSN) to be used to synchronize state information pertaining to an application among nodes of a state replication group (SRG) prior to a suspension of the SRG's operations is identified. Each node stores a respective commit record set of the application. Some number of SRG nodes suspend operations after synchronizing their local commit records up to the CSN. A configuration manager of the SRG verifies that, subsequent to a suspension of operations at the nodes, at least a threshold number of the nodes are available for service and have updated their commit record sets. The configuration manager then re-activates the SRG.

Abstract: During a training phase of a machine learning model, representations of at least some nodes of a decision tree are generated and stored on persistent storage in depth-first order. A respective predictive utility metric (PUM) value is determined for one or more nodes, indicating expected contributions of the nodes to a prediction of the model. A particular node is selected for removal from the tree based at least partly on its PUM value. A modified version of the tree, with the particular node removed, is stored for obtaining a prediction.

Abstract: A network-based virtual computing resource provider may offer virtual compute instances that implement rolling resource credits for scheduling virtual computing resources. Work requests for a virtual compute instance may be received at a virtualization manager. A resource credit balance may be determined for the virtual compute instance. The resource credit balance may accumulate resource credits in rolling fashion, carrying over unused credits from previous time periods. Resource credits may then be applied when generating scheduling instructions to provide to a physical resource to perform the work requests, such as a physical CPU in order to increase the utilization of the resource according to the number of credits applied. Applied resource credits may then be deducted from the credit balance.

Abstract: Data indicative of user proximity to one or more devices is provided to a server. Based on the proximity data, the server may modify content distribution to, or the behavior of, the one or more devices. In one implementation, determination that no user is proximate to the device may result in pausing or stopping streaming of content to the server.

Abstract: An application manager module provides anonymized user profile information to third party adaptive software applications. As a result, a software developer may produce a single software application that is adapted to run in a first mode providing full-functionality for use by adults and a second mode providing appropriate privacy and content restrictions for use by children. The mode is selected at run-time based on the anonymized user profile information received from the application manager module.

Abstract: In certain embodiments, a system comprises a memory and a processor communicatively coupled to the memory. The memory includes executable instructions that upon execution cause the system to generate, at a first time, a first snapshot capturing data stored in storage units of a storage device. The executable instructions upon execution cause the system to receive an indication to delete at least a portion of the data in the storage units and captured by the first snapshot, and to mark, in response to receiving the indication, the one or more storage units that store the at least a first portion of the data as available. The executable instructions upon execution cause the system to generate, at a second time subsequent to the first time, a second snapshot that omits the one or more storage units marked as available.

Abstract: An application programming interface for a data storage service provides a convenient mechanism for clients of the data storage service to access its various capabilities. An API call may be made to initiate a job and in response a job identifier may be provided. A separate API call specifying the job identifier may be made and a response providing information related to the job may result. Various API calls may be used to store data, retrieve data, obtain an inventory of stored data, and to obtain other information relating to stored data.

Abstract: Methods and systems for extending a bandwidth of a multi-band antenna of a user device are described. A multi-band antenna includes a single radio frequency (RF) input coupled to a first loop antenna, the first loop antenna configured to provide a first resonant mode. The multi-band antenna also includes a second antenna parasitically coupled to the first loop antenna to provide additional resonant modes of the multi-band antenna. The second antenna is a T-monopole antenna with a base coupled to the ground plane, a first arm extending out from a first side of the base, a second arm extending out from a second side of the base and a folded arm extending back towards the second side of the base from a distal end of the second arm.

Abstract: Methods, apparatus, and computer-accessible storage media for activating a gateway to a remote service provider. The gateway serves as an interface between processes on a customer network and the provider, for example to store customer data to a remote data store. A gateway sends a public key and metadata describing the gateway to the provider. The gateway receives an activation key from the provider and exposes the activation key on the customer network. The customer obtains the key and communicates to the provider using the key to provide customer information including a name for the gateway and to authorize registration of the gateway. The provider provides the customer information to the gateway. The gateway requests security credentials from the provider using the customer information and the key. The provider sends a security credential to the gateway. The gateway may then obtain configuration information from the customer via the provider.

Abstract: In speech processing systems, a special audio trigger indication is configured to efficiently isolate and mark incorrect speech processing results. The trigger indication may be configured to be easily recognizable by a speech processing device under various ASR and acoustic conditions. Once a speech processing device recognizes the trigger indication, incorrectly processed speech processing results are marked and may be isolated and prioritized for review by training and upgrading processes.

Abstract: A log-structured data store may implement optimized log storage for asynchronous log updates. In some embodiments, log records may be received indicating updates to data stored for a storage client and indicating positions in a log record sequence. The log records themselves may not be guaranteed to be received according to the log record sequence. Received log records may be stored in a hot log portion of a block-based storage device according to an order in which they are received. Log records in the hot log portion may then be identified to be moved to a cold log portion of the block-based storage device in order to complete a next portion of the log record sequence. Log records may be modified, such as compressed, or coalesced, before being stored together in a data block of the cold log portion according to the log record sequence.

Abstract: A caching system and methodology for data in a memory for faster access to commonly-used data by other applications and computer devices on a network. The memory can include a solid-state drive (SSD) array for the cache memory that has read-bias, in addition to a magnetic hard drive array. The system uses a logical set of slots to hold identifiers for specific groups of data that can be placed into cache memory and each identifier has a usage attribute that changes based upon the usage of the specific group of data and causes the identifier to move within the set of slots and potentially into cache memory.

Abstract: An electrowetting display device has a picture element having first and second support plates with a space in between, the space including first and second fluids immiscible with each other, the second fluid being electroconductive or polar. The first support plate includes a wall defining an extent of the picture element. An electric field is applied between an electrode and the second fluid, the electrode having an edge, a low portion and a high portion, the high portion being closer to the second support plate than the low portion, the high portion being arranged adjacent to the wall and having an area smaller than the area of the low portion. The high portion of the electrode has a cut-out causing an increased electric field in the space, and at least part of the edge of the electrode adjacent to the cut-out following the shape of the wall.

Abstract: A transcoding service is described that is capable of transcoding or otherwise processing content, such as video, audio or multimedia content, by utilizing one or more pipelines. A pipeline can enable a user to submit transcoding jobs (or other processing jobs) into an available pipeline, where a transcoding service (or other such service) assigns one or more computing resources to process the jobs received to each pipeline. The transcoding service and the pipelines can be provided by at least one service provider (e.g., a cloud computing provider) or other such entity to a plurality of customers. A service provider can also provide the computing resources (e.g., servers, virtual machines, etc.) used to process the transcoding jobs from the pipelines.

Abstract: Techniques are described for providing enhanced security for electronic communications, such as by including in a message sent between two services a digital signature that is generated by using secret information known to the services, so that the recipient receives assurance regarding the sender's identity if the recipient can replicate the received digital signature using the secret information known to the recipient. In some situations, the enhanced security is used in communications to and/or from an access manager system that provides single sign-on functionality and other functionality to other services for use with those services' users, such as to prevent malicious phishers from inappropriately gaining access to user information. Various services may use the enhanced security techniques when interacting with the access manager system at various times, such as to initiate sign-on for a user and/or to take subsequent action on behalf of a signed-on user.

Abstract: A user can capture various types of information concurrently using multiple sensors of an electronic device. This “scene” data can be provided to a service for processing, which is able to identify various types of potential matches and aggregate information to be returned to the client device. In at least some embodiments, matching information can be sent with the results such that the electronic device can match an element in the scene the next time that element is encountered, without having to contact the service again. In some embodiments, an attempt can be made to predict elements that the user might attempt to identify, and one or more corpora of data can be sent to the electronic device such that the client device can perform any matching on the device for elements in those corpora.

Abstract: Locating features in image information captured by multiple cameras of a computing device can be facilitated by removing background objects from the image information. A distance past which objects should be considered to be background objects is determined, and an amount of disparity for that distance calculated. Any objects having less than that amount of disparity in the image information can be removed, leaving objects that should be treated as foreground objects. An advantage of such an approach is that objects can be determined to be foreground objects even when those objects are only viewable by one of the cameras and disparity information cannot be calculated for those objects from the image information. Such an approach can be combined with other approaches as well to improve accuracy.

Abstract: An application may be installed on a user device. Installing the application may include receiving and storing an executable application, a plugin, and an application manifest. The application manifest may include a path corresponding to the plugin and one or more rules associated with the plugin. A webpage may be requested from the web server. The webpage may be parsed to determine that the webpage includes an instruction to use the plugin. It may be determined that the application includes the plugin. The path corresponding to the plugin may be obtained from the application manifest. The webpage may be presented using the plugin.

Abstract: Features are disclosed for discriminative training of speech recognition language models. A confusion matrix can be generated from acoustic model training data for use in discriminative training. The confusion matrix can include probabilities for the substitution, insertion, and/or deletion of some or all subword units of a language. Probabilities can be calculated based on the presence or absence of subword units in a processed acoustic model training data audio recording when compared to a correct transcription of the recording. The probabilities can be used to generate erroneous transcriptions in language model training corpora, and the language model can be trained to distinguish the erroneous transcriptions from the correct transcriptions.