Techniques for generating and outputting a natural language explanation of a determination made by a system are described. The system presents content to a user, where the content is generated based on a system determination. The system determines history data associated with a user profile associated with the user and context data associated with the system determination. The system uses the history data and the context data to determine a natural language explanation that the output was generated based on the system determination. The system further uses the history data and the context data to generate a predicted system determination representing the system determination that resulted in the output presented to the user. Based on a similarity between the predicted system determination and the actual system determination, the natural language explanation is presented to the user.
G10L 15/22 - Procédures utilisées pendant le processus de reconnaissance de la parole, p. ex. dialogue homme-machine
G06F 16/635 - Filtrage basé sur des données supplémentaires, p. ex. sur des profils d'utilisateurs ou de groupes
G10L 15/01 - Estimation ou évaluation des systèmes de reconnaissance de la parole
G10L 15/06 - Création de gabarits de référenceEntraînement des systèmes de reconnaissance de la parole, p. ex. adaptation aux caractéristiques de la voix du locuteur
G10L 15/18 - Classement ou recherche de la parole utilisant une modélisation du langage naturel
G10L 15/30 - Reconnaissance distribuée, p. ex. dans les systèmes client-serveur, pour les applications en téléphonie mobile ou réseaux
A natural language processing system may use system response configuration data to determine customized output data forms when outputting data for a user. The system response configuration data may represent various output attributes the system may use when creating output data. The system may have a certain number of existing profiles where a profile is associated with certain settings for the system response configuration data/attributes. The system may also use various data such as context data, sentiment data, or the like to customize system response configuration data during a dialog. Other components, such as natural language generation (NLG), text-to-speech (TTS), or the like, may use the customized system response configuration data to determine the form, timing, etc. of output data to be presented to a user.
G10L 13/047 - Architecture des synthétiseurs de parole
G10L 13/08 - Analyse de texte ou génération de paramètres pour la synthèse de la parole à partir de texte, p. ex. conversion graphème-phonème, génération de prosodie ou détermination de l'intonation ou de l'accent tonique
G10L 15/18 - Classement ou recherche de la parole utilisant une modélisation du langage naturel
G10L 15/22 - Procédures utilisées pendant le processus de reconnaissance de la parole, p. ex. dialogue homme-machine
3.
DISTRIBUTED DATABASE WITH INDEPENDENT SCALING OF COMMIT LAYER AND STORAGE LAYER
A database system includes a commit layer implemented using a first set of host computing devices and a storage layer implemented using a second set of host computing devices. A control plane of the distributed database system determines a first sharding scheme for the commit layer and a second sharding scheme for the storage layer, wherein the first and second sharding schemes are not required to be the same. Also, in some embodiments, the second sharding scheme used for the storage layer enables overlapping key spaces across the shards of the storage layer, wherein various ones of the shards are optimized for different types of workloads.
G06F 16/27 - Réplication, distribution ou synchronisation de données entre bases de données ou dans un système de bases de données distribuéesArchitectures de systèmes de bases de données distribuées à cet effet
4.
VERTICAL AND HORIZONTAL SCALING OF COMPONENTS OF A DISTRIBUTED DATABASE
A database system performs vertical scaling of a storage layer by temporarily increasing a resource allocation of given node and/or shard to allow the node or shard to process a load that exceeds its baseline resource allocation. Additionally, a control plane of the database system performs health checks of the nodes and/or shards of the components of the database system and in response to load conditions exceeding a threshold, performs horizontal scaling of the nodes of the components. The horizontal scaling adds shard replicas or re-shards the nodes to include more shards. The horizontal scaling reduces load on individual nodes and/or shards and alleviates the load conditions that triggered the vertical scaling.
G06F 16/27 - Réplication, distribution ou synchronisation de données entre bases de données ou dans un système de bases de données distribuéesArchitectures de systèmes de bases de données distribuées à cet effet
5.
REAL-TIME SEQUENTIAL CODE RECOMMENDATIONS WITH SYNTACTICALLY COMPLETE CODE COMPLETIONS
Disclosed are systems and methods that address the limitations of current code completion techniques, generate multiple levels of syntactically complete code completions, each level of syntactically complete code completion based upon and dependent upon an acceptance of a prior level syntactically complete code completion. A first level syntactically complete code completion may be presented as a suggestion for inclusion in a code and each additional level of syntactically complete code completions in the sequence maintained in a cache so that the next level syntactically complete code completion can be presented immediately upon acceptance of the currently presented syntactically complete code completion. By pre-generating multiple levels of syntactically complete code completions so that each next level syntactically complete code completion can be presented immediately upon acceptance of a presented syntactically complete code completion reduces or eliminates any perceived latency in code completion generation and/or code completion presentation.
A machine learning resource management service allows customers to define machine learning projects and machine learning resource allocations for the machine learning projects, such that different levels of resources are allocated to different ones of the projects. Additionally, the machine learning resource management service enables burst capacity at respective ones of the machine learning projects using under-utilized resources of other ones of the machine learning resources, while ensuring the customer defined resource allocations for the different machine learning projects are enforced. Additionally, the machine learning resource management service may track usage of burst capacity among the projects to ensure fair sharing of burst capacity.
G06F 9/50 - Allocation de ressources, p. ex. de l'unité centrale de traitement [UCT]
G06N 3/063 - Réalisation physique, c.-à-d. mise en œuvre matérielle de réseaux neuronaux, de neurones ou de parties de neurone utilisant des moyens électroniques
A database system may virtualize client connections to query processors to enable the query processors to be used by active connections rather than allowing the query processors to remain idle. Virtualizing the client connections may enable the database system and other systems sharing computing resources with the database system to operate with increased efficiency over a database system which does not virtualize client connections.
G06F 16/27 - Réplication, distribution ou synchronisation de données entre bases de données ou dans un système de bases de données distribuéesArchitectures de systèmes de bases de données distribuées à cet effet
8.
NONLINEAR TENSOR COMPRESSION AND DECOMPRESSION FOR NEURAL NETWORKS
Devices and techniques are generally described for nonlinear tensor compression for neural networks. In various examples, a first tensor associated with a first layer of a neural network may be determined. One or more neural processing units of accelerator hardware may generate a first compressed tensor by applying a nonlinear compression function to the first tensor. The first compressed tensor may be stored in a first memory of the one or more computer-readable media. A first operation associated with a second layer of the neural network may be determined, where the first operation uses output of the first layer. The first operation may be performed based on the first compressed tensor.
A database system performs vertical scaling of a storage layer by temporarily increasing a resource allocation of given node and/or shard to allow the node or shard to process a load that exceeds its baseline resource allocation. Additionally, a control plane of the database system performs health checks of the nodes and/or shards of the components of the database system and in response to load conditions exceeding a threshold, performs horizontal scaling of the nodes of the components. The horizontal scaling adds shard replicas or re-shards the nodes to include more shards. The horizontal scaling reduces load on individual nodes and/or shards and alleviates the load conditions that triggered the vertical scaling.
G06F 16/27 - Réplication, distribution ou synchronisation de données entre bases de données ou dans un système de bases de données distribuéesArchitectures de systèmes de bases de données distribuées à cet effet
G06F 9/50 - Allocation de ressources, p. ex. de l'unité centrale de traitement [UCT]
A database system provides query processors on demand for accepting customer connections to a database and stores database data in a separate storage layer, via storage nodes each storing a shard or shard replica of the database data. The database system provides a multi-region configuration wherein customers can access a multi-region database from any of multiple regions of a service provider network. In response to a region-wide failure event, query processors are provided on demand in a failover region. Additionally, to ensure sufficient storage node capacity is maintained in a potential failover region, a multi-region control plane distributes load or configuration information to local control planes of each of the regions of the multi-region database to ensure sufficient storage layer scaling is performed to support a failure over event resulting from a region-wide failure.
G06F 16/27 - Réplication, distribution ou synchronisation de données entre bases de données ou dans un système de bases de données distribuéesArchitectures de systèmes de bases de données distribuées à cet effet
11.
INTELLIGENT FILE SYSTEM WITH TRANSPARENT STORAGE TIERING
A file system manager implemented at a provider network identifies a storage device of a first group of storage devices of a provider network as an initial location of a file system object. Based on an access metric associated with the object, the file system manager initiates a transfer of contents of the object to a second storage device of a different storage device group, without receiving a client request specifying the transfer. In response to an access request received via a file system programmatic interface, contents of the object are provided from the second storage device. Based on a second access metric, the object is transferred back to the first group of storage devices.
G06F 3/06 - Entrée numérique à partir de, ou sortie numérique vers des supports d'enregistrement
G06F 16/11 - Administration des systèmes de fichiers, p. ex. détails de l’archivage ou d’instantanés
G06F 16/185 - Systèmes de gestion de stockage hiérarchisé, p. ex. migration de fichiers ou politiques de migration de fichiers
G06Q 20/10 - Architectures de paiement spécialement adaptées aux systèmes de transfert électronique de fondsArchitectures de paiement spécialement adaptées aux systèmes de banque à domicile
12.
DISTRIBUTED DATABASE WITH INDEPENDENT SCALING OF COMMIT LAYER AND STORAGE LAYER
A database system includes a commit layer implemented using a first set of host computing devices and a storage layer implemented using a second set of host computing devices. A control plane of the distributed database system determines a first sharding scheme for the commit layer and a second sharding scheme for the storage layer, wherein the first and second sharding schemes are not required to be the same. Also, in some embodiments, the second sharding scheme used for the storage layer enables overlapping key spaces across the shards of the storage layer, wherein various ones of the shards are optimized for different types of workloads.
Techniques for moderating an output of a generative model in a streaming manner are described. In some embodiments, a first portion of data (responsive to an input) may be generated by a generative model, a system may process the first portion of data using a content moderation model to determine that the first portion corresponds to a non-moderated content category, and based on this determination, the first portion of data may be outputted (to a user or system component). The generative model may then generate a second portion of data (which may include a larger of number tokens than the second portion), and the system may process the second portion using the content moderation model to determine whether the second portion corresponds to a moderated content category. The amount of data (e.g., number of tokens) processed by the content moderation model may vary between processing steps.
Approaches are disclosed for providing (412) optimized AI models for use in performing various inferencing tasks. In at least one embodiment, a user may request a model to be used to perform an inferencing task, and may be presented (406) with one or more optimization options. The user can select (408) one or more of these optimization options, and in response a model and parameter set can be provided (410) to the user, where the model and/or parameter set may be optimized and/or proprietary, and thus have their use restricted. Such an approach allows a user to effectively obtain a customized AI model that can be used for a specific type of inferencing task without the need to fine-tune or customize the model. In order to protect any intellectual property (IP), such as an optimized parameter set offered by a provider, the set may be encrypted and able to be decrypted and used (614) only in authorized environments and associated (616) with users having a valid key or cryptographic token associated with the set of optimized parameters.
A modular system (e.g., for establishing circulation availability of liquid coolant for datacenter components) can include a set of cabinets couplable together to form a coolant loop having a supply side and a return side. The cabinets can include at least one pressure imparting cabinet, at least one coolant distributing cabinet, and/or at least one heat exchanging cabinet. A pump included in a pressure imparting cabinet may circulate coolant through the coolant loop. A manifold included in a coolant distributing cabinet may distribute coolant along the supply side of the coolant loop toward heat-generating components and direct coolant carrying heat from said components into the return side of the coolant loop. A heat exchanger included in a heat exchanging cabinet may be arranged for dissipating heat carried in the coolant loop so as to ready the coolant for use along the supply side.
Approaches presented herein relate to an answer refinement system that may be included as part of a generative artificial intelligence (AI) pipeline. As content is produced by one or more generative AI models, the answer refinement system may segment the answer into chunks and then validate information within each of the chunks. Chunks that include invalid information may be rewritten or otherwise modified to correct errors. Chunks that are valid may be further analyzed for conditional validity and conditionally valid chunks may be modified to provide further context or assumptions for validity.
Techniques for connecting to cloud-hosted instances without requiring those instances to have a public network address are described. A first WebSocket message including a first payload is received from an electronic device via a WebSocket connection. A first TCP/IP message including at least a portion of the first payload is sent to an instance hosted by a cloud provider network, the instance having a first network address on a first virtual network, and the first TCP/IP message including a second network address as a source address, traffic originating from the second network address being routable to the first virtual network. A second TCP/IP message including a second payload is received from the instance, the second TCP/IP message including the second network address as a destination address. A second WebSocket message including at least a portion of the second payload is sent to the electronic device sending via the WebSocket connection.
H04L 69/16 - Implémentation ou adaptation du protocole Internet [IP], du protocole de contrôle de transmission [TCP] ou du protocole datagramme utilisateur [UDP]
As a media program is aired to listeners, a control system monitors audio data transmitted to the listeners and interactions received from the listeners to determine whether the media program has violated or may violate one or more rules. The audio data is processed to identify words expressed therein and features of the audio data. Additionally, features of users (e.g., a creator or any listeners or guests) may be calculated based on any information or data available regarding such users. An embedding is formed with data representing the words, the audio features and the user features, and provided to a model trained to determine whether a media program is at risk of violating any rules. One or more actions are selected and executed or recommended based on a score generated by the model representing a level of risk that a rule has been, is being or will be violated.
G10L 15/16 - Classement ou recherche de la parole utilisant des réseaux neuronaux artificiels
H04H 60/65 - Dispositions pour des services utilisant les résultats du contrôle, de l'identification ou de la reconnaissance, couverts par les groupes ou pour utiliser les résultats côté utilisateurs
Systems and methods for account association with voice-enabled devices are disclosed. For example, a voice-enabled device situated in a managed environment, such as a hotel room, may be taken by a temporary resident or guest of the environment. Upon determining that the device has been removed from the environment, a device identifier associated with the device may be dissociated from components and/or services associated with environment and/or systems related thereto, and the device identifier may be associated with a user account of the user.
Devices, systems, and methods for enhancing user privacy by using anonymized delivery labels may include identifying, by a first device, a computer-readable code on a parcel to be delivered to a delivery address, wherein delivery information of the parcel is absent from the parcel; sending, by the first device, a unique identifier of the parcel included in the computer-readable code to a second device that has pre-authenticated to a third device associated with maintaining delivery information for packages; sending, by the second device, the unique identifier to the third device; determining, by the third device, based on receiving the unique identifier, that delivery information criteria for the parcel are satisfied; sending, by the third device, the delivery information to the second device based on determining that the delivery information criteria for the parcel are satisfied; and causing presentation of the delivery information.
G06K 7/14 - Méthodes ou dispositions pour la lecture de supports d'enregistrement par radiation électromagnétique, p. ex. lecture optiqueMéthodes ou dispositions pour la lecture de supports d'enregistrement par radiation corpusculaire utilisant la lumière sans sélection des longueurs d'onde, p. ex. lecture de la lumière blanche réfléchie
G01C 21/34 - Recherche d'itinéraireGuidage en matière d'itinéraire
G03B 21/00 - Projecteurs ou visionneuses du type par projectionLeurs accessoires
G06F 21/44 - Authentification de programme ou de dispositif
Techniques for reducing a syndrome density of a plurality of rounds of syndrome measurements following a first decoding stage (e.g., via a local decoder) for quantum error correction of circuit-level noise within quantum surface codes are disclosed. Such techniques for reducing syndrome density may include syndrome collapse and/or vertical cleanup techniques. In a syndrome collapse technique, a measurement results volume may be partitioned into sheets and the respective sheets collapsed, causing vertical pairs of highlighted vertices to be removed. In a vertical cleanup technique, vertical pairs of highlighted vertices may be removed directly from a matching graph following a first decoding stage. Following the removal of vertical pairs of highlighted vertices, the measurement results are then decoded in a second, global decoding stage. Such techniques allow for fast decoding throughout and low latency times for error correction of rounds of syndrome measurements for quantum algorithms implemented using quantum surface codes.
G06N 10/80 - Programmation quantique, p. ex. interfaces, langages ou boîtes à outils de développement logiciel pour la création ou la manipulation de programmes capables de fonctionner sur des ordinateurs quantiquesPlate-formes pour la simulation ou l’accès aux ordinateurs quantiques, p. ex. informatique quantique en nuage
Technologies of a device-based Fast Presence Detection (FPD) for a contactless sleep-tracking device are described. One method of a sleep-monitoring device includes receiving radar data from a radar unit. The radar data includes i) first data representing a breathing waveform associated with a user, ii) a first set of range values, and iii) a first set of confidence values associated with the first data. The method determines absolute magnitude values, first infinite impulse response (IIR) values using the first set of range values, and second IIR values using the first set of confidence values. The method determines a first event representing the user located in a first region using the absolute magnitude values and the first and second IIR values. The method sends an indication of the first event to a cloud service that causes one or more devices in the environment to perform one or more actions.
First participant information may be received that is associated with a set of participants that participate in an event. The set of participants may be distributed, based at least in part on distribution criteria and the first participant information, across a plurality of messaging groups, to form a first participant distribution, wherein each messaging group of the plurality of messaging groups has a respective participant subset of the set of participants, and wherein messages sent by participants within the respective participant subset are delivered only to other participants within the respective participant subset. During the event, second participant information may be received associated with the set of participants. Also during the event, the first participant distribution may be modified, based at least in part on the distribution criteria and the second participant information, to form a modified participant distribution.
Techniques for user interface defect detection in media player applications are described. According to some embodiments, a computer-implemented method includes receiving a request at a cloud provider network to perform a defect detection on a media player application, capturing an image of a user interaction with a user interface of the media player application, determining, by the cloud provider network, one or more components of the user interface from pixels of the image, detecting, by the cloud provider network, a defect in the user interface from the one or more components without creating a reference image, and generating, by the cloud provider network, an output based at least in part on the defect.
G06F 3/0484 - Techniques d’interaction fondées sur les interfaces utilisateur graphiques [GUI] pour la commande de fonctions ou d’opérations spécifiques, p. ex. sélection ou transformation d’un objet, d’une image ou d’un élément de texte affiché, détermination d’une valeur de paramètre ou sélection d’une plage de valeurs
G06F 3/0481 - Techniques d’interaction fondées sur les interfaces utilisateur graphiques [GUI] fondées sur des propriétés spécifiques de l’objet d’interaction affiché ou sur un environnement basé sur les métaphores, p. ex. interaction avec des éléments du bureau telles les fenêtres ou les icônes, ou avec l’aide d’un curseur changeant de comportement ou d’aspect
Systems and methods for LM communication channel optimization include receiving user input data requesting that a message be sent and determining, using a language model (LM), a recipient profile to send the message to. Thereafter, the LM may query a communication channel application for data indicating communication channels available for sending the message to the recipient profile, and then the LM may infer, based on the data, an urgency value and/or formality value to associate with the message. In this example, the LM may be trained to infer the urgency value and/or the formality value from content of the message. Then, a communication channel may be selected based at least in part on the urgency value and/or the formality value.
Systems and method for real-time keywords recommendations are provided. The systems and methods leverage one or more machine learning models that receive information about events that will occur in the future. The one or more machine learning models perform parallel processing to determine, in real-time and before the events occur, different keywords that are likely to experience an increase in usage based on the events. The one or more machine learning models also determine different types of content produced by content originators that are relevant to the determined keywords. Recommendations may be made for the content originator to have an association performed between the keywords and the content. Once the associations are performed, when a consumer inputs the keyword into an application, the consumer may be presented with the content or a mechanism (such as a hyperlink, for example) by which the consumer may access the content.
A wireless charging device for charging a head-mounted wearable device (HMWD) includes a base, a sidewall that extends from the base, and a bridge support that extends from the base and is spaced apart from the sidewall. At least one charging antenna is positioned within the sidewall. The HMWD is engaged with the charging device by placing the nose bridge of the HMWD in contact with the top of the bridge support, while the temples of the HMWD extend into the space between the sidewall and bridge support. The sidewall, bridge support, and base constrain movement of the temples relative to the charging device in a manner that retains the receiving antennae in the temples within a range of positions relative to the charging antenna that are suitable to receive electrical power.
H02J 50/00 - Circuits ou systèmes pour l'alimentation ou la distribution sans fil d'énergie électrique
H02J 50/90 - Circuits ou systèmes pour l'alimentation ou la distribution sans fil d'énergie électrique mettant en œuvre la détection ou l'optimisation de la position, p. ex. de l'alignement
Systems and methods are described for implementing debugging snapshots on a serverless computing system. A serverless computing system executes user-submitted code in sandboxed execution environments such as virtual machines or containers, and the user who requests execution of the code does not have direct access to these execution environments for debugging or other purposes. To support debugging of code, the serverless computing system thus implements a debugging snapshot service that generates snapshots of the environment in which the user-submitted code is executing. Snapshots are generated accordance with criteria that may be specified by the user, and may include any or all of the information needed to resume execution of the code from the point at which the snapshot was taken. The service includes user interfaces that enable inspection and comparison of snapshots, as well as setting snapshot generation and retention policies.
G06F 9/455 - ÉmulationInterprétationSimulation de logiciel, p. ex. virtualisation ou émulation des moteurs d’exécution d’applications ou de systèmes d’exploitation
29.
Secure connectivity from external clients to dynamically changing cloud resource groups
An endpoint for accessing a group of cloud resources from a set of client devices outside the cloud is established. In response to detecting that, as a result of a configuration change, a particular cloud resource has joined the group, addressing information for the particular cloud resource is generated. An access verifier associated with the endpoint receives a packet directed from a client device using the addressing information. In response to determining, based on user identity metadata of the user and based on device status metadata of the client device, that the packet satisfies a security requirement, the packet is delivered to the particular cloud resource.
G06F 21/00 - Dispositions de sécurité pour protéger les calculateurs, leurs composants, les programmes ou les données contre une activité non autorisée
Systems and methods are provided for efficiently building an object detection learning model for an unlabeled pool of images. A recommendation engine automatically recommends an annotation type for the images in the unlabeled pool based on previous object detection and an updated mean average precision of the model, where the mean average precision represents the performance of the model.
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
G06V 10/774 - Génération d'ensembles de motifs de formationTraitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source méthodes de Bootstrap, p. ex. "bagging” ou “boosting”
G06V 10/776 - ValidationÉvaluation des performances
G06V 10/778 - Apprentissage de profils actif, p. ex. apprentissage en ligne des caractéristiques d’images ou de vidéos
Techniques for reducing occurrences of cross-triggering event types not represented in audio data and false detection of event types are described. Different event types, such as a hand clap event type and a door knock event type may have substantially similar audio characteristics, and if one event type of such event types is represented in audio data, then event detection processing of that audio data may lead to detection of event types not represented in the audio data. Example embodiments involve training a model configured to detect multiple event types to enforce mutual exclusivity between different event type pairs or sets of the multiple event types. The model is trained to enforce mutual exclusivity using a regularizer function and a weight parameter to reduce any positive detection scores of event types not represented in received audio. Similar techniques may be applied to models for object detection using image data.
G10L 15/06 - Création de gabarits de référenceEntraînement des systèmes de reconnaissance de la parole, p. ex. adaptation aux caractéristiques de la voix du locuteur
G10L 15/32 - Reconnaisseurs multiples utilisés en séquence ou en parallèleSystèmes de combinaison de score à cet effet, p. ex. systèmes de vote
32.
Replaceable interconnect cartridge with handle and guide for top installation
A midplane frame may be contained in a rack-mountable enclosure and define at least a first bay. Guides may be distributed among a first cartridge and the midplane frame and arranged to facilitate aligned vertical movement of the first cartridge along a height direction into a landed position in the first bay. The first cartridge may have forwardly-oriented connectors arrayed in rows arranged in a stack in a height direction. A plurality of appliances may each have a row of one or more rearwardly-oriented connectors, and each of the appliances may be movable rearwardly along a length direction in the enclosure into a seated arrangement in which the appliances are stacked over one another in the height direction and in which the rows of rearwardly-oriented connectors of the appliances are coupled with the rows of the forwardly-oriented connectors of the first cartridge in the landed position.
Systems and techniques are disclosed for upsampling low resolution images in remote sensing data, such as satellite images, into higher-resolution upsampled images. A machine learning upsampling model is trained on a training data set containing crowdsourced high resolution images, such as dashcam images, cell phone camera images, and other types of images of geographical areas, as well as corresponding low resolution images from remote sensing data that depict the same geographical areas. The upsampling model is trained on the training data set to determine an upsampling approach that converts the low resolution images into upsampled images that match the crowdsourced high resolution images of the same geographical areas. Following training of the upsampling model, the upsampling model is used to upsample new low resolution images in remote sensing data into higher-resolution upsampled images.
Embodiments of a supply chain management system (SCMS) are disclosed that enable the generation of synthetic supply chain activity data for developing machine learning models, such as models for predicting vendor lead times (VLTs) of purchase orders fulfilled by a supply chain network. In embodiments, the generation process is performed over successive time periods to simulate dynamically changing variables of the supply chain network, including inventory levels, product demand, and stock manager decisions. The generation process may also be used to generate synthetic data to simulate elements within the supply chain network, such as simulated warehouses, vendors, or products. The disclosed SCMS is able to generate highly realistic training data that simulates the operations within the supply chain network, which can be used to improve the performance of machine learning models.
Techniques for prompt template optimization with language models are described. In some examples, a prompt template optimization request to optimize a generative artificial intelligence model prompt template is received, the prompt template optimization request including an initial prompt template and an indication of a selected function, the selected function to implement at least a portion of a prompt template optimization workflow. The prompt template optimization workflow is processed with the selected function, the prompt template optimization workflow including one or more iterations of generating, evaluating, and selecting prompt template variants based at least in part on the initial prompt template to yield a final prompt template. The final prompt template is output.
G06F 16/383 - Recherche caractérisée par l’utilisation de métadonnées, p. ex. de métadonnées ne provenant pas du contenu ou de métadonnées générées manuellement utilisant des métadonnées provenant automatiquement du contenu
G06F 3/0482 - Interaction avec des listes d’éléments sélectionnables, p. ex. des menus
36.
Modular thread analytics exploration for extrapolating reasons from complex database
A graphical user interface receives natural language input from a user. A modular thread analytics exploration system uses context determination, dynamic context enrichment, and the natural language input to generate a solution recipe with a language model. The system prompt the language model with evaluation guides to improve the accuracy of the model output. The solution recipe includes steps (i) that are used to generate code and (ii) that are used to generate natural language explanations. The system generates code with a language model. The system processes the generated code in a sandbox and self-debugs the generated code as necessary. The output from the steps is presented in the graphical user interface.
Implementations are described herein for unlocking a wireless device using image analysis and liveliness detection. A wireless device may capture, using a camera of the wireless device, an image of a person that is interacting with the wireless device. The wireless device may transmit a first signal using a first antenna and may receive a second signal using a second antenna. The wireless device may determine whether the image corresponds to a stored image. The wireless device may determine whether the second signal indicates a movement of the person or a depth characteristic of the person. The wireless device may selectively unlock the wireless device based on whether the image matches a stored image of the plurality of the stored images and based on whether the second signal indicates at least one of the movement of the person or the depth characteristic of the person.
G06F 21/32 - Authentification de l’utilisateur par données biométriques, p. ex. empreintes digitales, balayages de l’iris ou empreintes vocales
G01S 13/28 - Systèmes pour mesurer la distance uniquement utilisant la transmission de trains discontinus d'ondes modulées par impulsions dans lesquels les impulsions émises utilisent une onde porteuse modulée en fréquence ou en phase avec compression dans le temps des impulsions reçues
G01S 13/88 - Radar ou systèmes analogues, spécialement adaptés pour des applications spécifiques
G06F 21/44 - Authentification de programme ou de dispositif
G06V 10/70 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique
G06V 40/16 - Visages humains, p. ex. parties du visage, croquis ou expressions
G06V 40/40 - Détection d’usurpation, p. ex. détection d’activité
38.
Determining actions associated with communications in a multi-channel artificial intelligence architecture
Systems and methods for multi-channel Artificial Intelligence (AI) architectures include receiving data representing a communication, such as a document. A format associated with the document may be determined. Once the format associated with the document is determined, a preprocessing model configured to process data associated with the format may be used with the data to generate text data representing the document. A first portion of the text data may be identified from the text data. A processing model may then be used to determine an action associated with the document based at least in part on the first portion of the text data. An application programming interface (API) may then be selected to send a request to for executing the action. The document may also be associated with a user account of the user such that the user may subsequently request information that may be included in the document from various devices.
H04L 51/224 - Surveillance ou traitement des messages en fournissant une notification sur les messages entrants, p. ex. des poussées de notifications des messages reçus
39.
Customer-specified routing option groups and selection policies for cloud network traffic
A traffic manager obtains (a) a representation of an association between a set of networking destinations and a routing option group, and (b) a policy for selecting routing options from the group for network packets. For a network packet directed to one of the destinations, the traffic manager selects one of the routing options of the group based on the policy, and causes the packet to be transmitted to the destination along a path. The path includes, as a next-hop address, a network address associated with the selected routing option.
Described are example systems and methods generally directed to determining a location score in connection with a geographic area of interest that may represent a suitability of the geographic area of interest in connection with the performance of a service. A geographic area of interest is divided into a plurality of cells and one or more constraints in connection with an area of interest (or combination of multiple areas of interest, etc.) is determined, a mapping function is defined for each constraint, and the constraints in determining a location score for each cell of the area of interest is aggregated. In exemplary implementations, the location score for each cell of the area of interest may represent and/or correspond to a suitability of the area of interest in connection with performing aerial deliveries of items using an aerial vehicle.
This disclosure is directed to an item-identifying, mobile cart that may be utilized by a user in a materials handling facility to automatically identify a user operating the cart and items that the user places into a basket of the cart. In addition, the cart may update a virtual shopping cart of the identified user to include items taken by the user. The mobile cart may include multiple imaging devices and oriented such that their respective optical axes are directed towards an interior of a perimeter of the top of the basket, and above the top of the basket. The mobile cart may also include an imaging device oriented away from the basket such that a user operating the mobile cart may scan a user identifier using this imaging device to enable recognition of the user.
B62B 3/14 - Voitures à bras ayant plus d'un essieu portant les roues servant au déplacementDispositifs de direction à cet effetAppareillage à cet effet caractérisées par des moyens pour l'emboîtement ou l'empilage, p. ex. chariots pour achats
B62B 5/00 - Accessoires ou détails spécialement adaptés aux voitures à bras
G06V 20/52 - Activités de surveillance ou de suivi, p. ex. pour la reconnaissance d’objets suspects
Systems, methods, and computer-readable media are disclosed for wireless charging of industrial equipment. In one embodiment, an example system may include a first mat configured to wirelessly charge a first device and a second device, the first mat having a first charging coil disposed in a first region of the first mat, and a second charging coil disposed in a second region of the first mat. The system may include a controller configured to determine, at a first time, that the first device is in contact with the first region of the first mat, and cause the first charging coil to be energized for wireless charging of the first device.
H02J 7/00 - Circuits pour la charge ou la dépolarisation des batteries ou pour alimenter des charges par des batteries
H02J 50/00 - Circuits ou systèmes pour l'alimentation ou la distribution sans fil d'énergie électrique
H02J 50/10 - Circuits ou systèmes pour l'alimentation ou la distribution sans fil d'énergie électrique utilisant un couplage inductif
H02J 50/40 - Circuits ou systèmes pour l'alimentation ou la distribution sans fil d'énergie électrique utilisant plusieurs dispositifs de transmission ou de réception
H02J 50/90 - Circuits ou systèmes pour l'alimentation ou la distribution sans fil d'énergie électrique mettant en œuvre la détection ou l'optimisation de la position, p. ex. de l'alignement
A computer-implemented method includes generating or receiving instruction code for executing by a computing device to implement a neural network model, where the instruction code includes a plurality of direct memory access (DMA) instructions for data transferring between a local memory of an accelerator of the computing device and a system memory of the computing device; modifying the instruction code to arrange sources or destinations of a group of DMA instructions of the plurality of DMA instructions into a contiguous block in the local memory; and replacing the group of DMA instructions with a single DMA instruction, wherein a source address or a destination address of the single DMA instruction is the contiguous block of the local memory.
Techniques for enabling dynamic secondary content insertion in multiple view (multiview) video streaming using bitstream stitching techniques are described. According to some examples, a computer-implemented method includes sending a first live video stream and a second live video stream having a same group of pictures duration to a single decoder of a device for simultaneous viewing; receiving an indication of a break within a group of pictures of the first live video stream for displaying a secondary content video stream; sending, in response to the receiving the indication, one or more fill frames to the single decoder of the device to display between a start of the break and an end of the group of pictures of the first live video stream for simultaneous viewing with the second live stream; and sending, in response to the receiving the indication, the secondary content video stream having the same group of pictures duration as the first live video stream to the single decoder of the device for simultaneous viewing with the second live stream after displaying of the one or more fill frames.
H04N 21/2662 - Contrôle de la complexité du flux vidéo, p. ex. en mettant à l'échelle la résolution ou le débit binaire du flux vidéo en fonction des capacités du client
46.
Systolic array with output rounding across multiple data streams
Systems and methods are provided to round the numbers produced by a systolic array. A rounder can receive a number from the systolic array and identify a data stream associated with the number from a plurality of data streams. The rounder can identify a random number generator. The random number generator may be associated with a random number sequence and may generate a next random number in the random number sequence based on a state value representing a position within the random number sequence. The data stream may be associated with a respective state value representing a current position for the data stream. Based on the current position for the data stream, the rounder can initialize a state value of the random number generator. The rounder can perform a rounding operation using the initialized state value of the random number generator.
G06F 7/499 - Maniement de valeur ou d'exception, p. ex. arrondi ou dépassement
G06F 7/483 - Calculs avec des nombres représentés par une combinaison non linéaire de nombres codés, p. ex. nombres rationnels, système de numération logarithmique ou nombres à virgule flottante
G06F 15/80 - Architectures de calculateurs universels à programmes enregistrés comprenant un ensemble d'unités de traitement à commande commune, p. ex. plusieurs processeurs de données à instruction unique
47.
Multi-modal, reconfigurable, and adaptive gripping system and method to handle item variability
A multi-modal, reconfigurable, and adaptive gripper system may include a suction cup assembly, a static finger assembly, and at least two reconfigurable finger assemblies that are controlled by a pressure-regulated actuation assembly. In order to grasp an item, a grasp mode, a finger configuration, and/or force(s) to apply to the item may be selected or determined. Various combinations of the suction cup assembly and finger assemblies may be used, with various finger configurations, and with various air pressures or differentials supplied by the pressure-regulated actuation assembly, in order to apply the selected force(s) to portions of the item and reliably grasp, transport, and release the item as part of various automated material handling processes.
09 - Appareils et instruments scientifiques et électriques
42 - Services scientifiques, technologiques et industriels, recherche et conception
Produits et services
Computer hardware for executing and accelerating machine learning inference workloads; computer hardware for deploying and running trained machine learning models in production environments; computer hardware for high-throughput, low-latency execution of artificial intelligence (AI) and machine learning inference; downloadable computer software for using a trained AI model to make predictions, answer prompts, generate text, images, and video, or analyze new data; downloadable computer software for performance monitoring, profiling, and debugging of machine learning model training and inference; machine learning (ML) accelerator chips; artificial intelligence (AI) accelerator processors; all of the foregoing for use with custom machine learning chips. Providing temporary use of on-line non-downloadable cloud computing software for executing and accelerating machine learning inference workloads; Providing temporary use of on-line non-downloadable cloud computing software for deploying and running trained machine learning models in production environments; Providing temporary use of on-line non-downloadable cloud computing software for high-throughput, low-latency execution of artificial intelligence (AI) and machine learning inference; Providing temporary use of on-line non-downloadable cloud computing software for performance monitoring, profiling, and debugging of machine learning model training and inference; Providing temporary use of on-line non-downloadable cloud computing software for using a trained AI model to make predictions, answer prompts, generate text, images, and video, or analyze new data; Technical consulting and support services in the field of custom AI hardware; Advising others on optimizing machine learning workloads using specialized chips; all of the foregoing for use with custom machine learning chips.
(1) Industrial robots; humanoid robots; programmable robots; robotic platforms for research and development; service robots; industrial robots for handling, palletizing and movement of goods and workpieces; cargo handling robots; robotic arms for industrial use; robot component parts; robot component accessories
(2) Humanoid robots having communication and learning functions for assisting and entertaining people; user-programmable humanoid robots, not configured; downloadable software for monitoring and controlling communication between computers and automated machine systems; downloadable operating system software for robots; downloadable software for machine learning for use in robots; downloadable software for programming physical movements for use in robots; downloadable software using artificial intelligence (AI) for speech recognition for use in robots; downloadable software development kits (SDK); security surveillance robots; humanoid robots with artificial intelligence for use in entertainment; education; scientific research; preparing beverages; assisting human beings with household cleaning and laundry; assisting humans in trade fairs; assisting humans in museum and exhibition tour guides; assisting human beings with household chores and tasks; assisting humans in concierge duties and tasks; assisting humans in business management of logistics; taking customer orders and serving and collecting dishes in restaurants; humanoid robots with artificial intelligence for use in providing physical labor and recreational activity, companionship, and real time information and analysis; supporting operations in manufacturing, logistics, warehousing, and retail settings, namely, performing inventory management, transporting goods, restocking shelves, and assisting customers; humanoid robots with artificial intelligence for use in industrial settings, namely, performing repetitive assembly tasks, quality control inspections, and hazardous material handling; character-based experiences; retail associate experiences; event-based experiential marketing; downloadable computer simulation software for modeling humanoid robot behavior and humanoid robot environments
(3) Computers; application programming interface (API) software; networking software; wireless communication devices for voice or data transmission; humanoid robots with artificial intelligence; user-programmable humanoid robots; telepresence robots; laboratory robots and structural and replacement parts therefor; configurable humanoid robots for manipulation of objects in human environments for the purposes of logistics and warehousing, featuring pre-installed operating software, and structural and replacement parts therefor; humanoid robots with artificial intelligence for manipulation of objects in human environments, namely, for retrieving, carrying and moving containers in warehouses; humanoid robotic components, namely, humanoid robotics platforms in the nature of robots for personal, educational and hobby use and structural parts therefor; humanoid robotic components, namely, robotic arms for laboratory purposes; tactical robots; downloadable application programming interface (API) software for robotics software development and device integration; downloadable software for controlling, operating and programming robots; downloadable software for perception, navigation and autonomous operation of robots; downloadable software using artificial intelligence for simulating natural conversation; downloadable computer chatbot software for simulating conversations (1) Electronic transmission of data, commands, and instructions to robots and automated systems; rental of telepresence robots; telecommunications services for remotely controlling, monitoring, and operating robots and automated machines; provision of access to networks for communication between robots, humans, and automated systems; telepresence and remote operation services for humanoid robots and autonomous devices
(2) Rental of humanoid robots with artificial intelligence (AI); design and development of software; design and development of computer hardware; design and development of new products; technical consulting in the field of monitoring technological functions of humanoid robots with artificial intelligence (AI); rental of user-programmable humanoid robots, not configured; technical support services, namely, troubleshooting of computer software problems; providing temporary use of online non-downloadable software development kits (SDKs); providing online non-downloadable virtual assistant software using artificial intelligence (AI) for answering customer inquiries, scheduling appointments, assisting with reservations, managing tasks, providing information and guidance in hospitality, retail, education, and event environments, supporting instructional and teaching activities; software as a service (SAAS) services featuring software for the design, development, operation and management of humanoid robots, programmable humanoid robots, humanoid robotic devices, and components thereof; platform as a service (PAAS) featuring computer software platforms for the design, development, operation and management of humanoid robots, programmable humanoid robots, humanoid robotic devices, and components thereof; scientific and technological services, namely, research and design in the field of humanoid robots and humanoid robotic systems; product design and development in the field of humanoid robots and humanoid robotic systems; design and development of software and hardware for operating and maneuvering humanoid robots and humanoid robotic systems; providing temporary use of online non-downloadable chatbot software for simulating conversations, providing customer assistance, answering questions, giving directions, delivering concierge-style interactions, offering educational and instructional dialogue, supporting learning experiences, enabling character-based conversational interactions, facilitating developer testing of interactive humanoid robotic behaviors, and supporting teleoperated interactions through conversational interfaces; providing subscription-based temporary use of on-line non-downloadable software for operating and maneuvering humanoid robots and humanoid robotic systems; providing online non-downloadable virtual assistant software using artificial intelligence (AI) for performing personal assistant functions, enabling character-based interactive experiences, assisting developers and researchers in building, testing, and demonstrating humanoid robotic applications, and supporting teleoperation workflows for remotely controlled humanoid robotic behaviors; providing temporary use of online non-downloadable simulation software for modeling humanoid robot behavior and humanoid robot environments
(3) Computer consultation, testing, programming, advisory, and maintenance services; research and development of computer hardware, software, and artificial intelligence; leasing computers, computer peripherals, software, and AI systems; software consulting; support and consultation for developing computer systems, databases, and applications; providing online information about computer software and hardware; creating indexes of information, searching and retrieving information; troubleshooting, installation, and maintenance of computer software; computer security services; monitoring and remote monitoring of computer systems; software as a service (SaaS) and platform as a service (PaaS) for artificial intelligence, machine learning, natural language processing, robotics; software engineering and computer software development; design and development of software and hardware for operating, maneuvering, and controlling robots and robotic systems; technical consulting and research in the field of humanoid robots with artificial intelligence; providing online non-downloadable virtual assistant software using artificial intelligence (AI) for assisting developers and researchers in building, testing, and demonstrating humanoid robotic applications, and supporting teleoperation workflows for remotely controlled humanoid robotic behaviors; computer software consulting and computer programming services
(1) Computer hardware for executing and accelerating machine learning inference workloads; computer hardware for deploying and running trained machine learning models in production environments; computer hardware for high-throughput, low-latency execution of artificial intelligence (AI) and machine learning inference; downloadable computer software for using a trained AI model to make predictions, answer prompts, generate text, images, and video, or analyze new data; downloadable computer software for performance monitoring, profiling, and debugging of machine learning model training and inference; machine learning (ML) accelerator chips; artificial intelligence (AI) accelerator processors; all of the foregoing for use with custom machine learning chips (1) Providing temporary use of on-line non-downloadable cloud computing software for executing and accelerating machine learning inference workloads; Providing temporary use of on-line non-downloadable cloud computing software for deploying and running trained machine learning models in production environments; Providing temporary use of on-line non-downloadable cloud computing software for high-throughput, low-latency execution of artificial intelligence (AI) and machine learning inference; Providing temporary use of on-line non-downloadable cloud computing software for performance monitoring, profiling, and debugging of machine learning model training and inference; Providing temporary use of on-line non-downloadable cloud computing software for using a trained AI model to make predictions, answer prompts, generate text, images, and video, or analyze new data; Technical consulting and support services in the field of custom AI hardware; Advising others on optimizing machine learning workloads using specialized chips; all of the foregoing for use with custom machine learning chips
09 - Appareils et instruments scientifiques et électriques
38 - Services de télécommunications
42 - Services scientifiques, technologiques et industriels, recherche et conception
Produits et services
Industrial robots; humanoid robots; programmable robots; robotic platforms for research and development; service robots; industrial robots for handling, palletizing and movement of goods and workpieces; cargo handling robots; robotic arms for industrial use; robot component parts; robot component accessories. Humanoid robots having communication and learning functions for assisting and entertaining people; user-programmable humanoid robots, not configured; downloadable software for monitoring and controlling communication between computers and automated machine systems; downloadable operating system software for robots; downloadable software for machine learning for use in robots; downloadable software for programming physical movements for use in robots; downloadable software using artificial intelligence (AI) for speech recognition for use in robots; downloadable software development kits (SDK); security surveillance robots; humanoid robots with artificial intelligence for use in entertainment; education; scientific research; preparing beverages; assisting human beings with household cleaning and laundry; assisting humans in trade fairs; assisting humans in museum and exhibition tour guides; assisting human beings with household chores and tasks; assisting humans in concierge duties and tasks; assisting humans in business management of logistics; taking customer orders and serving and collecting dishes in restaurants; humanoid robots with artificial intelligence for use in providing physical labor and recreational activity, companionship, and real time information and analysis; supporting operations in manufacturing, logistics, warehousing, and retail settings, namely, performing inventory management, transporting goods, restocking shelves, and assisting customers; humanoid robots with artificial intelligence for use in industrial settings, namely, performing repetitive assembly tasks, quality control inspections, and hazardous material handling; character-based experiences; retail associate experiences; event-based experiential marketing; downloadable computer simulation software for modeling humanoid robot behavior and humanoid robot environments; computers; application programming interface (API) software; networking software; wireless communication devices for voice or data transmission; humanoid robots with artificial intelligence; user-programmable humanoid robots; telepresence robots; laboratory robots and structural and replacement parts therefor; configurable humanoid robots for manipulation of objects in human environments for the purposes of logistics and warehousing, featuring pre-installed operating software, and structural and replacement parts therefor; humanoid robots with artificial intelligence for manipulation of objects in human environments, namely, for retrieving, carrying and moving containers in warehouses; humanoid robotic components, namely, humanoid robotics platforms in the nature of robots for personal, educational and hobby use and structural parts therefor; humanoid robotic components, namely, robotic arms for laboratory purposes; tactical robots; downloadable application programming interface (API) software for robotics software development and device integration; downloadable software for controlling, operating and programming robots; downloadable software for perception, navigation and autonomous operation of robots; downloadable software using artificial intelligence for simulating natural conversation; downloadable computer chatbot software for simulating conversations. Electronic transmission of data, commands, and instructions to robots and automated systems; rental of telepresence robots; telecommunications services for remotely controlling, monitoring, and operating robots and automated machines; provision of access to networks for communication between robots, humans, and automated systems; telepresence and remote operation services for humanoid robots and autonomous devices. Rental of humanoid robots with artificial intelligence (AI); design and development of software; design and development of computer hardware; design and development of new products; technical consulting in the field of monitoring technological functions of humanoid robots with artificial intelligence (AI); rental of user-programmable humanoid robots, not configured; technical support services, namely, troubleshooting of computer software problems; providing temporary use of online non-downloadable software development kits (SDKs); providing online non-downloadable virtual assistant software using artificial intelligence (AI) for answering customer inquiries, scheduling appointments, assisting with reservations, managing tasks, providing information and guidance in hospitality, retail, education, and event environments, supporting instructional and teaching activities; software as a service (SaaS) services featuring software for the design, development, operation and management of humanoid robots, programmable humanoid robots, humanoid robotic devices, and components thereof; platform as a service (PaaS) featuring computer software platforms for the design, development, operation and management of humanoid robots, programmable humanoid robots, humanoid robotic devices, and components thereof; Scientific and technological services, namely, research and design in the field of humanoid robots and humanoid robotic systems; product design and development in the field of humanoid robots and humanoid robotic systems; design and development of software and hardware for operating and maneuvering humanoid robots and humanoid robotic systems; providing temporary use of online non-downloadable chatbot software for simulating conversations, providing customer assistance, answering questions, giving directions, delivering concierge-style interactions, offering educational and instructional dialogue, supporting learning experiences, enabling character-based conversational interactions, facilitating developer testing of interactive humanoid robotic behaviors, and supporting teleoperated interactions through conversational interfaces; providing subscription-based temporary use of on-line non-downloadable software for operating and maneuvering humanoid robots and humanoid robotic systems; providing online non-downloadable virtual assistant software using artificial intelligence (AI) for performing personal assistant functions, enabling character-based interactive experiences, assisting developers and researchers in building, testing, and demonstrating humanoid robotic applications, and supporting teleoperation workflows for remotely controlled humanoid robotic behaviors; providing temporary use of online non-downloadable simulation software for modeling humanoid robot behavior and humanoid robot environments; computer consultation, testing, programming, advisory, and maintenance services; research and development of computer hardware, software, and artificial intelligence; leasing computers, computer peripherals, software, and AI systems; software consulting; support and consultation for developing computer systems, databases, and applications; providing online information about computer software and hardware; creating indexes of information, searching and retrieving information; troubleshooting, installation, and maintenance of computer software; computer security services; monitoring and remote monitoring of computer systems; software as a service (SaaS) and platform as a service (PaaS) for artificial intelligence, machine learning, natural language processing, robotics; software engineering and computer software development; design and development of software and hardware for operating, maneuvering, and controlling robots and robotic systems; technical consulting and research in the field of humanoid robots with artificial intelligence; providing online non-downloadable virtual assistant software using artificial intelligence (AI) for assisting developers and researchers in building, testing, and demonstrating humanoid robotic applications, and supporting teleoperation workflows for remotely controlled humanoid robotic behaviors; computer software consulting and computer programming services.
A content broadcast system may allow a user to select and start an audio stream of desired audio content without having to connect and authenticate to a specific device. Rather than a user having to pause the content and reconfigure settings of the broadcast system to select the desired audio content, the system may broadcast advertisements listing available audio content (e.g., corresponding to different spoken languages) and actively listen for requests from a device for new audio content to be streamed with the content. A user may manually select the new audio content, or the listening device may request particular audio content based on user preferences (e.g., a preferred language for streaming content). The system may broadcast audio data using a Bluetooth protocol.
H04N 21/442 - Surveillance de procédés ou de ressources, p. ex. détection de la défaillance d'un dispositif d'enregistrement, surveillance de la bande passante sur la voie descendante, du nombre de visualisations d'un film, de l'espace de stockage disponible dans le disque dur interne
53.
NATURAL LANGUAGE INTERACTIONS USING VISUAL UNDERSTANDING
Techniques for performing an action with respect to displayed content are described. A natural language interpretation corresponding to a received spoken user input may be determined. Prior to receiving the spoken user input, content may be displayed to the user from which the spoken user input was received. The natural language interpretation may represent a request to perform an action with respect to a portion of the content currently being displayed. Content identifiers corresponding to content being displayed, may be determined, and embedding data representing at least one feature of the content may be determined using the content identifiers. The natural language interpretation and the embedding data may be processed to determine that the spoken user input relates to a first portion of the displayed content instead of a second portion of the displayed content. Based on the determination, an action responsive to the spoken user input may be performed.
G10L 15/18 - Classement ou recherche de la parole utilisant une modélisation du langage naturel
G10L 15/19 - Contexte grammatical, p. ex. désambiguïsation des hypothèses de reconnaissance par application des règles de séquence de mots
G10L 15/24 - Reconnaissance de la parole utilisant des caractéristiques non acoustiques
G10L 15/30 - Reconnaissance distribuée, p. ex. dans les systèmes client-serveur, pour les applications en téléphonie mobile ou réseaux
G10L 25/57 - Techniques d'analyse de la parole ou de la voix qui ne se limitent pas à un seul des groupes spécialement adaptées pour un usage particulier pour comparaison ou différentiation pour le traitement des signaux vidéo
54.
HYBRID SUCTION END OF ARM TOOLS HAVING DYNAMICALLY VARIABLE SUCTION ARRAYS
Systems and methods are disclosed for hybrid suction end of arm tools having dynamically variable suction arrays and related item manipulation devices. In one embodiment, an example item manipulation device may include a housing, a first suction cup assembly having a first suction cup and a first suction cup support arm, where the first suction cup support arm is configured to rotate with respect to the housing, and a second suction cup assembly having a second suction cup and a second suction cup support arm, where the second suction cup support arm is configured to rotate with respect to the housing. At least one of the first suction cup assembly and the second suction cup assembly can be configured to move relative to the other.
Techniques for moderating an output of a generative model in a streaming manner are described. In some embodiments, a first portion of data (responsive to an input) may be generated by a generative model, a system may process the first portion of data using a content moderation model to determine that the first portion corresponds to a non-moderated content category, and based on this determination, the first portion of data may be outputted (to a user or system component). The generative model may then generate a second portion of data (which may include a larger of number tokens than the second portion), and the system may process the second portion using the content moderation model to determine whether the second portion corresponds to a moderated content category. The amount of data (e.g., number of tokens) processed by the content moderation model may vary between processing steps.
Approaches presented herein relate to an answer refinement system that may be included as part of a generative artificial intelligence (AI) pipeline. As content is produced by one or more generative AI models, the answer refinement system may segment the answer into chunks and then validate information within each of the chunks. Chunks that include invalid information may be rewritten or otherwise modified to correct errors. Chunks that are valid may be further analyzed for conditional validity and conditionally valid chunks may be modified to provide further context or assumptions for validity.
G06F 16/215 - Amélioration de la qualité des donnéesNettoyage des données, p. ex. déduplication, suppression des entrées non valides ou correction des erreurs typographiques
G06F 16/3329 - Formulation de requêtes en langage naturel
A machine learning resource management service allows customers to define machine learning projects and machine learning resource allocations for the machine learning projects, such that different levels of resources are allocated to different ones of the projects. Additionally, the machine learning resource management service enables burst capacity at respective ones of the machine learning projects using under-utilized resources of other ones of the machine learning resources, while ensuring the customer defined resource allocations for the different machine learning projects are enforced. Additionally, the machine learning resource management service may track usage of burst capacity among the projects to ensure fair sharing of burst capacity.
A modular system (e.g., for establishing circulation availability of liquid coolant for datacenter components) can include a set of cabinets couplable together to form a coolant loop having a supply side and a return side. The cabinets can include at least one pressure imparting cabinet, at least one coolant distributing cabinet, and/or at least one heat exchanging cabinet. A pump included in a pressure imparting cabinet may circulate coolant through the coolant loop. A manifold included in a coolant distributing cabinet may distribute coolant along the supply side of the coolant loop toward heat-generating components and direct coolant carrying heat from said components into the return side of the coolant loop. A heat exchanger included in a heat exchanging cabinet may be arranged for dissipating heat carried in the coolant loop so as to ready the coolant for use along the supply side.
Disclosed are systems and methods that address the limitations of current code completion techniques, generate multiple levels of syntactically complete code completions, each level of syntactically complete code completion based upon and dependent upon an acceptance of a prior level syntactically complete code completion. A first level syntactically complete code completion may be presented as a suggestion for inclusion in a code and each additional level of syntactically complete code completions in the sequence maintained in a cache so that the next level syntactically complete code completion can be presented immediately upon acceptance of the currently presented syntactically complete code completion. By pre-generating multiple levels of syntactically complete code completions so that each next level syntactically complete code completion can be presented immediately upon acceptance of a presented syntactically complete code completion reduces or eliminates any perceived latency in code completion generation and/or code completion presentation.
Approaches are disclosed for providing optimized AI models for use in performing various inferencing tasks. In at least one embodiment, a user may request a model to be used to perform an inferencing task, and may be presented with one or more optimization options. The user can select one or more of these optimization options, and in response a model and parameter set can be provided to the user, where the model and/or parameter set may be optimized and/or proprietary, and thus have their use restricted. Such an approach allows a user to effectively obtain a customized AI model that can be used for a specific type of inferencing task without the need to fine-tune or customize the model. In order to protect any intellectual property (IP), such as an optimized parameter set offered by a provider, the set may be encrypted and able to be decrypted and used only in authorized environments and associated with users having a valid key or cryptographic token associated with the set of optimized parameters.
Systems and methods are disclosed for active and passive electromagnetic switching for sortation shuttles along a track. An example system for active and passive electromagnetic switching for sortation shuttles may include a track having a first linear path and a first curved path that intersects the first linear path. The system may include a shuttle with a first ferrous block, the shuttle configured to move along the track, a first set of electromagnets disposed along a side of the first curved path, and a first set of permanent magnets disposed along a side of the first linear path. Energizing the first set of electromagnets causes the shuttle to merge onto the first curved path via interaction with the first ferrous block.
Systems are generally described that include a curved light guide for thin structure illumination. An example system includes a light sub-assembly comprising a curved light sub-assembly backing ring and a plurality of light-emitting diodes (LEDs), each LED of the plurality of LEDs being coupled to the curved light sub-assembly backing ring. The example system also includes a curved light guide having an edge coupled to the light sub-assembly, the curved light guide including a pattern of optical extraction features that distribute light and are positioned on the exterior surface of the curved light guide for uniformly distributing light from the plurality of LEDs. The example system also includes a curved reflector including an exterior surface coupled to an interior surface of the curved light guide, wherein the exterior surface is reflective, and a volumetric diffuser coupled to the exterior surface of the curved light guide.
F21V 14/06 - Commande de la distribution de la lumière émise par réglage d’éléments constitutifs par un mouvement de réfracteurs
F21K 9/232 - Sources lumineuses rétrocompatibles pour dispositifs d’éclairage avec un seul culot pour chaque source lumineuse, p. ex. pour le remplacement de lampes à incandescence avec un culot à baïonnette ou à vis spécialement adaptées à la génération de lumière essentiellement omnidirectionnelle, p. ex. avec une ampoule en verre
F21K 9/61 - Agencements optiques intégrés dans la source lumineuse, p. ex. pour améliorer l’indice de rendu des couleurs ou l’extraction de lumière en utilisant des guides de lumière
F21K 9/66 - Détails des globes ou des couvercles faisant partie de la source lumineuse
F21S 8/04 - Dispositifs d'éclairage destinés à des installations fixes destinés uniquement au montage sur un plafond ou sur une structure similaire en porte-à-faux
F21V 3/04 - GlobesVasquesVerres de protection caractérisés par les matériaux, traitements de surface ou revêtements
F21V 3/06 - GlobesVasquesVerres de protection caractérisés par les matériaux, traitements de surface ou revêtements caractérisés par le matériau
F21V 8/00 - Utilisation de guides de lumière, p. ex. dispositifs à fibres optiques, dans les dispositifs ou systèmes d'éclairage
F21V 21/34 - Éléments de support déplaçables le long d'un élément de guidage
F21V 21/35 - Éléments de support déplaçables le long d'un élément de guidage avec un contact électrique direct entre l'élément de support et les conducteurs électriques disposés le long de l'élément de guidage
F21V 33/00 - Combinaisons structurales de dispositifs d'éclairage avec d'autres objets, non prévues ailleurs
F21Y 103/33 - Sources lumineuses de forme allongée, p. ex. tubes fluorescents courbes annulaires
F21Y 105/18 - Sources lumineuses planes comprenant un réseau bidimensionnel d’éléments générateurs de lumière ponctuelle caractérisées par la forme d’ensemble du réseau bidimensionnel annulaireSources lumineuses planes comprenant un réseau bidimensionnel d’éléments générateurs de lumière ponctuelle caractérisées par la forme d’ensemble du réseau bidimensionnel polygonale autre que rectangulaire ou carrée, p. ex. pour les spots lumineux ou pour générer un faisceau lumineux axialement symétrique
G08B 13/196 - Déclenchement influencé par la chaleur, la lumière, ou les radiations de longueur d'onde plus courteDéclenchement par introduction de sources de chaleur, de lumière, ou de radiations de longueur d'onde plus courte utilisant des systèmes détecteurs de radiations passifs utilisant des systèmes de balayage et de comparaison d'image utilisant des caméras de télévision
64.
Quantum key distribution network management service
A system and method enabling a management service to dynamically select a key relay technique between at least a first relay technique that uses more quantum key distribution (QKD) bits and a second relay technique that uses less QKD key bits and select a path for relaying a key between a source QKD node and a destination QKD node. Respective QKD nodes may relay information about QKD key bit inventory to the management service, wherein the management service may store respective data in a repository. Management service may receive a request for distribution of a QKD key and select one or more key relay techniques to relay the key at respective QKD node links. Additionally, the management service may dynamically select and optimize the relay path and the key relay technique for respective links based on QKD key bit information.
Variations in latency, out-of-order, and duplication may occur for incoming packets delivered via a network including a constellation of low-Earth orbit (LEO) satellites. An incoming packet that comprises time data and a sequence number is received at a user terminal. A delivery deadline time (deadline) is determined for the incoming packet. The incoming packet and its deadline are stored in a waiting buffer. Packets from the waiting buffer are processed for storage into “slots” that correspond to sequence numbers of the incoming packets. A window designates which portion of the slots may be written to or read from. The window may comprise a circular buffer. The window may be “moved” relative to the slots based on sequence number of an incoming packet, highest packet transmitted, maximum permitted movement, lowest window stop, highest window stop, and so forth. Packets in slots within the window that have reached their deadline are sent.
H04L 67/12 - Protocoles spécialement adaptés aux environnements propriétaires ou de mise en réseau pour un usage spécial, p. ex. les réseaux médicaux, les réseaux de capteurs, les réseaux dans les véhicules ou les réseaux de mesure à distance
Techniques implementable by a computer system are provided. The techniques include sending a request to stream media content. The request can include a media content identifier and a streaming start point in the media content. The techniques also include receiving an encrypted portion of a media stream for the media content. The encrypted portion can be encrypted by an encryption key. The portion can begin at a silence point. The silence point can be at or after a threshold time length beyond the streaming start point. The techniques also include receiving the encryption key. The techniques also include presenting the encrypted portion of the media stream.
H04N 21/2347 - Traitement de flux vidéo élémentaires, p. ex. raccordement de flux vidéo ou transformation de graphes de scènes du flux vidéo codé impliquant le cryptage de flux vidéo
H04N 21/233 - Traitement de flux audio élémentaires
H04N 21/239 - Interfaçage de la voie montante du réseau de transmission, p. ex. établissement de priorité des requêtes de clients
H04N 21/254 - Gestion au sein du serveur de données additionnelles, p. ex. serveur d'achat ou serveur de gestion de droits
H04N 21/845 - Structuration du contenu, p. ex. décomposition du contenu en segments temporels
Systems and methods are disclosed for contactless direction of sortation shuttles along a track. An example system for contactless direction of sortation shuttles may include a track having a linear path, and a curved path that intersects the linear path. The system may include a shuttle with a first ferrous block and a second ferrous block, the shuttle configured to move along the track, and a first set of electromagnets disposed along a side of the curved path. Electromagnets of the first set of electromagnets may be configured to be individually energized. Energizing the first set of electromagnets may cause the shuttle to merge onto the curved path via interaction with at least one of the first ferrous block or the second ferrous block.
B60L 13/00 - Propulsion électrique pour véhicules à monorail, véhicules suspendus ou chemins de fer à crémaillèreSuspension ou lévitation magnétiques pour véhicules
B60L 13/08 - Moyens pour déterminer ou commander la position ou l'assiette du véhicule relativement à la voie pour la position latérale
B61B 13/12 - Systèmes avec dispositifs de propulsion entre les rails ou le long de ceux-ci, p. ex. systèmes pneumatiques
B65G 35/06 - Transporteurs mécaniques non prévus ailleurs comportant un porte-charges se déplaçant le long d'un circuit, p. ex. d'un circuit fermé, et adapté pour venir en prise avec l'un quelconque des éléments de traction espacés le long du circuit
B65G 54/02 - Transporteurs non mécaniques, non prévus ailleurs électrostatiques, électriques ou magnétiques
B60L 13/03 - Propulsion électrique par moteur linéaire
B65G 1/137 - Dispositifs d'emmagasinage mécaniques avec des aménagements ou des moyens de commande automatique pour choisir les objets qui doivent être enlevés
H02K 41/03 - Moteurs synchronesMoteurs pas à pasMoteurs à réluctance
69.
Virtual machine host health monitoring with untrusted sources in a cloud provider network
Techniques for monitoring virtual machine host system health with untrusted sources are described. An agent receives a request to terminate a first virtual machine, the request including an untrusted status indicator originating from an environment executing untrusted software. The agent sends first termination event data to a differential health service of the provider network, the first termination event data including an indication of a host computer system and the untrusted status indicator. The differential health service determines that a first metric associated with the first host computer system differs from a second metric associated with a pool of host computer systems by at least a first amount and based at least in part on the untrusted status indicator, wherein the pool of host computer systems includes the first host computer system. The differential health service sends a second request to cause a corrective action to be taken.
G06F 9/50 - Allocation de ressources, p. ex. de l'unité centrale de traitement [UCT]
G06F 21/54 - Contrôle des utilisateurs, des programmes ou des dispositifs de préservation de l’intégrité des plates-formes, p. ex. des processeurs, des micrologiciels ou des systèmes d’exploitation au stade de l’exécution du programme, p. ex. intégrité de la pile, débordement de tampon ou prévention d'effacement involontaire de données par ajout de routines ou d’objets de sécurité aux programmes
G06F 21/55 - Détection d’intrusion locale ou mise en œuvre de contre-mesures
Techniques for log storage in distributed data streaming systems are described. A cluster of brokers receive log records from publishers and send log records to subscribers. The log is represented as a group of segments, each segment subdivided into chunks. Metadata describes the log structure. Log records are stored in chunks at least in a remote storage location shared amongst the brokers in the cluster.
Techniques to verify the connectivity between standard cell pins and via ladder structures can include obtaining a mapping of standard cell pins to respective via ladders' characteristics, and determining a sufficient number of connections for connecting a standard cell pin to a corresponding via ladder in an integrated circuit design based on the via ladder characteristics of the standard cell pin. A data structure that models the integrated circuit design as a resistance-capacitance network is used to verify that the standard cell pin has the sufficient number of connections to the corresponding via ladder.
G06F 30/398 - Vérification ou optimisation de la conception, p. ex. par vérification des règles de conception [DRC], vérification de correspondance entre géométrie et schéma [LVS] ou par les méthodes à éléments finis [MEF]
G06F 30/392 - Conception de plans ou d’agencements, p. ex. partitionnement ou positionnement
Some embodiments provide a neural network inference circuit (NNIC) for executing a network having multiple layers. The NNIC includes multiple circuit sets. Each circuit set includes a dot product circuit to compute dot products between weight values and activation values for at least a subset of a first set of the layers, a math function circuit to compute values based on computations using activation values for at least a subset of a second set of the layers, and a post-processing circuit to receive (i) values output by the dot product circuit and (ii) values output by the math function circuit and to perform post-processing operations on the received values. The NNIC includes a set of accumulation circuits. Each accumulation circuit is to accumulate outputs of math function circuits for layers of the second set of layers that perform matrix multiplication of sets of activation values output by previous layers.
G06N 3/063 - Réalisation physique, c.-à-d. mise en œuvre matérielle de réseaux neuronaux, de neurones ou de parties de neurone utilisant des moyens électroniques
A token-based biometric payment processing system facilitates point-of-sale (POS) processing of payments. A biometric identification provider (BIP) receives a single-use token (SUT) from a payment processor (PP). The SUT is associated at the PP with a payment account on file (PAOF) and with an identity at the BIP. Upon identification at a scanner by the BIP, the SUT is processed to generate payment data (PMT) that is secured. The PMT is sent to the scanner, which forwards the PMT to a PP terminal using a communication interface and protocol. The communication interface and protocol may use existing contactless approaches, such as those used for mass transit cards, allowing the scanner to communicate with a wide array of PP terminals. The PP terminal then sends the PMT on to the PP. The PP receives the PMT and, if the PMT is valid, initiates a transaction to transfer funds from the PAOF.
G06Q 20/40 - Autorisation, p. ex. identification du payeur ou du bénéficiaire, vérification des références du client ou du magasinExamen et approbation des payeurs, p. ex. contrôle des lignes de crédit ou des listes négatives
H04L 9/32 - Dispositions pour les communications secrètes ou protégéesProtocoles réseaux de sécurité comprenant des moyens pour vérifier l'identité ou l'autorisation d'un utilisateur du système
74.
Machine learning techniques for content delivery service
A system is disclosed to host a first machine learning model and a second machine learning model, where the first machine learning model is trained using a first privacy-compliant dataset, and the second machine learning model is trained using a second privacy-compliant dataset. The system is to obtain a digital content request that includes a dataset including user data and digital content data. The dataset is processed using the first machine learning model to generate a first score representing a probability of a user engaging with digital content and using the second machine learning model to generate a second score representing a similar probability. Based on one or both of these scores, the system selects at least one digital content from a plurality of digital content.
This disclosure describes systems and techniques for recognizing, or identifying, produce items being purchased at a self-checkout device. The self-checkout device may use a produce-recognition system that uses image-based recognition techniques so that items, such as produce, may be identified without the need for users to enter in keywords or price look-up (PLU) codes. The produce-recognition system may compare feature representations between the produce to be purchased, produce categories, and produce subcategories. Based on similarities with produce categories and produce subcategories, the identity of the produce item may be predicted. Ranked produce categories and subcategories may presented to the user. The presentation of produce categories and subcategories may depend on confidence levels associated with the predicted produce categories. The produce item may then be identified and added to a virtual cart.
G06Q 20/18 - Architectures de paiement impliquant des terminaux en libre-service, des distributeurs automatiques, des bornes ou des terminaux multimédia
G06Q 20/20 - Systèmes de réseaux présents sur les points de vente
G06V 10/74 - Appariement de motifs d’image ou de vidéoMesures de proximité dans les espaces de caractéristiques
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
G06V 10/94 - Architectures logicielles ou matérielles spécialement adaptées à la compréhension d’images ou de vidéos
G06V 20/52 - Activités de surveillance ou de suivi, p. ex. pour la reconnaissance d’objets suspects
Devices and techniques are described for embedding-free speaker diarization. In some examples, a first speaker ID label is determined for a first frame and a second speaker ID label may be determined for a second frame of a first window of audio. A third speaker ID label may be determined for a third frame of a second window. First combined data representing at least the first frame and the third frame and second combined data representing at least the second frame and the third frame may be generated. First posterior data associated with the first frame and second posterior data associated with the third frame may be generated. Third posterior data associated with the second frame and fourth posterior data associated with the third frame may be generated. A determination may be made that the first speaker ID label and the third speaker ID label correspond to the same speaker.
G10L 21/028 - Séparation du signal de voix utilisant les propriétés des sources sonores
G10L 17/02 - Opérations de prétraitement, p. ex. sélection de segmentReprésentation ou modélisation de motifs, p. ex. fondée sur l’analyse linéaire discriminante [LDA] ou les composantes principalesSélection ou extraction des caractéristiques
In various examples, systems and methods of dynamic channel selection for wireless communication in the presence of interference. In various examples, first interference data for a first gateway device may be determined. In some cases, a wireless transmission scheme may be switched from a static wireless transmission scheme to a frequency hopping wireless transmission scheme based on the first interference data. In some examples, the first gateway device may send first instructions indicating that the first gateway device is switching to the frequency hopping wireless transmission scheme to a first remote computing device.
Techniques for a hybrid network directory service are described. Messages forming a request to launch an instance within a cloud provider network are received, the messages including an identifier of a customer virtual network within the cloud provider network, the customer virtual network having connectivity to another network outside of the cloud provider network, the other network outside of the cloud provider network having a directory service. An instance is launched, the instance having connectivity to the customer virtual network. A server of the directory service on the other network outside of the cloud provider network is identified. The identified server is caused to add the instance as a node of the directory service. Directory service data received from the identified server is stored. Directory service requests originating from the customer virtual network are processed.
Systems and methods are described for wakeup sensor threshold autocalibration. An example method includes generating sensor data by a wakeup sensor. The example method also includes determining that the sensor data represents a potential motion event based on a comparison of the sensor data with a sensor threshold value. The example method also includes controlling a camera device to capture video data and generating, by an onboard motion verification model and based on the video data, motion verification data labeling the potential motion event as a true positive motion event. The example method also includes determining an updated sensor threshold value by modifying the sensor threshold value by a first amount, where the first amount is determined based on the first motion verification data.
Disclosed are various embodiments that provide multipath connectivity between edge locations and regions in a cloud provider network. In one embodiment, data is received at a provider substrate extension of a cloud provider network, where the data is to be transmitted to a destination in a data center of the cloud provider network. The provider substrate extension is at an edge location of the cloud provider network. The data is transmitted to a destination in the data center of the cloud provider network via a plurality of communication links from the edge location to the data center.
H04W 40/12 - Sélection d'itinéraire ou de voie de communication, p. ex. routage basé sur l'énergie disponible ou le chemin le plus court sur la base de la qualité d'émission ou de la qualité des canaux
H04W 28/02 - Gestion du trafic, p. ex. régulation de flux ou d'encombrement
41 - Éducation, divertissements, activités sportives et culturelles
Produits et services
Entertainment in the nature of an ongoing television series in the fields of comedy and drama; entertainment services, namely, an ongoing television program in the fields of comedy and drama provided through television, cable, the Internet and wireless communications networks
09 - Appareils et instruments scientifiques et électriques
Produits et services
Pre-recorded downloadable audio recordings featuring comedic and dramatic entertainment programs; pre-recorded video recordings featuring comedic and dramatic entertainment programs; pre-recorded downloadable audio and visual recordings featuring comedic and dramatic entertainment programs
09 - Appareils et instruments scientifiques et électriques
Produits et services
Pre-recorded downloadable audio recordings featuring dramatic entertainment programs; pre-recorded video recordings featuring dramatic entertainment programs; pre-recorded downloadable audio and visual recordings featuring dramatic entertainment programs
41 - Éducation, divertissements, activités sportives et culturelles
Produits et services
Entertainment in the nature of an ongoing television series in the field of drama; entertainment services, namely, an ongoing television program in the field of drama provided through television, cable, the Internet and wireless communications networks
A system configured to improve SSL processing and/or target goal detection by fusing SSL data with object information to generate a combined target likelihood estimate that takes into account what the device knows about the surrounding environment. For example, the device may generate object information by performing object detection, floorplan estimation, distance measurements, and/or the like. Using this object information, the device may calculate a likelihood estimate value for each direction around the device, with known objects (e.g., walls) corresponding to low likelihood values. In response to an acoustic event (e.g., wakeword detection), the device may fuse the target likelihood estimates generated using SSL data and/or object information to generate the combined target likelihood estimate. Thus, the combined target likelihood estimate enables the device to accurately associate the acoustic event with a corresponding SSL track (e.g., direct sound) and ignore reflections caused by objects in the environment.
Systems and methods are disclosed for identifying whether an object is stored within a container (e.g., tote). The system generates, using two or more neural networks, a plurality of predictions on whether the object is stored in a container. The two or more neural networks use two or more sets of images that are captured within different time frames to generate the plurality of predictions. Then, the system determines whether the object is stored in a container based, at least in part, on the plurality of predictions.
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
Techniques for determining one or more expected categories of data for receipt by a skill or application and comparing the expected categories of data to user disclosed data, such as from a spoken natural language user input. The techniques include determining the spoken natural language user input is directed at an interaction with a particular skill and identifying a set of data categories registered for receipt by the skill. The techniques may further include determining if the skill solicited the disclosed data of the user input or if the user mistakenly provided the disclosed data. In some embodiments, the interaction with the skill may end if the skill is not authorized to receive the disclosed data. In some embodiments, the unsolicited disclosed data may be used to identify misunderstood or confusing requests for user input.
An Identity and Access Management Service implements persistent source values PSVs) for assumed identities. A source value (e.g., an original identifier of an entity) is persisted across assumed identities, facilitating identification of entities (users or applications) responsible for actions taken by the assumed (e.g., alternative) identities. The Manager receives a request to assume an identity. The request includes the entities current credentials and a PSV. The current credentials are authenticated and a persistent source value policy may be relied on to determine whether and/or how to grant the assumed identity. The PSV may be copied from credentials in the request in order to be included in the credentials for the requested identity that the Manager provides in response to the request. Use of the requested credentials, including the PSV, to access services or resources may be logged, the logs including the PSV from the request to assume the identity.
Systems and techniques are described for providing video enhancements during sports broadcasts. In various examples, first tracking data representing first respective locations of a first plurality of players at a first time may be received. First embedding data representing a formation of the first plurality of players at the first time may be generated based at least in part on the first tracking data. A first defensive coverage may be predicted using the first embedding data based at least in part on a similarity between the first respective locations of the first plurality of players at the first time and second respective locations of a second plurality of players in a historical play. A first graphical overlay may be displayed on a live video feed, where the first graphical overlay indicating the first defensive coverage.
Systems and methods are disclosed for a shuttle sortation system having a calibration shuttle for determining and adjusting shuttle coordinates for depositing items and/or loading totes at various tote receiving areas. The shuttle sortation system may include item tracks and tote tracks for sorting and/or distributing items, packages, and/or totes, and may move the shuttles about the shuttle sortation system via linear synchronous motors (LSMs). The calibration shuttle may include a base portion designed to interface with the track (e.g., with the LSMs). The base may support imaging systems having cameras and processors for capturing images of the tote receiving areas and processing the images to determine a centerline of such tote receiving areas and adjusting coordinates used to navigate the shuttle to the tote receiving areas. The calibration shuttle may be navigated to each tote receiving area to adjust coordinates where an offset from the centerline is detected.
Smart camera systems may comprise one or more modular assemblies, including a camera assembly, a light assembly, and a compute assembly. Each of the modular assemblies may be further formed from various modular components, circuits, or elements, and also include various modular software, applications, or algorithms related to image capture and processing. Using the modular assemblies, various different smart camera systems may be assembled that are adapted for different applications or environments.
G03B 17/14 - Corps d'appareils avec moyens pour supporter des objectifs, des lentilles additionnelles, des filtres, des masques ou des tourelles de façon interchangeable
G03B 15/03 - Combinaisons d'appareils photographiques avec appareils d'éclairageFlash
G03B 17/55 - Parties constitutives des appareils ou corps d'appareilsLeurs accessoires avec des dispositions pour chauffer ou réfrigérer, p. ex. avion
H04N 23/52 - Éléments optimisant le fonctionnement du capteur d'images, p. ex. pour la protection contre les interférences électromagnétiques [EMI] ou la commande de la température par des éléments de transfert de chaleur ou de refroidissement
H04N 23/54 - Montage de tubes analyseurs, de capteurs d'images électroniques, de bobines de déviation ou de focalisation
H04N 23/56 - Caméras ou modules de caméras comprenant des capteurs d'images électroniquesLeur commande munis de moyens d'éclairage
H04N 23/90 - Agencement de caméras ou de modules de caméras, p. ex. de plusieurs caméras dans des studios de télévision ou des stades de sport
92.
Techniques to determine region route guiding points using convergence circles
Techniques for generating a route for a computer-guided vehicle are disclosed. A server device can receive a graph with an ordered connection of regions of a map of a space. The computer system can determine a start point and an end point in the graph that define a subset of the points in the graph. The computer system can then generate, using the graph, a route by designating a current guide point, determining a convergence radius for each point in the subset of the points, and evaluating a visibility criterion between the current guide point and the subset of the points. Based on a point failing the visibility criterion, the computer system can designate an additional point that passed the visibility criterion as the current guide point, add the current guide point to the route, and iterate until both the start point and the end point are in the route.
G05D 1/00 - Commande de la position, du cap, de l'altitude ou de l'attitude des véhicules terrestres, aquatiques, aériens ou spatiaux, p. ex. utilisant des pilotes automatiques
G01C 21/36 - Dispositions d'entrée/sortie pour des calculateurs embarqués
93.
Forecasting wind conditions for operations of uncrewed aerial vehicles
Grids representing predictions of wind vectors (or velocities) over a ground region are obtained from a model and downscaled to a greater level of resolution based on surface roughness metrics of the ground region. Observations of wind vectors over the ground region are determined from ground-based or airborne systems and assimilated into the grids. Subsequently, grids representing the assimilated wind vectors at an initial time, and the predictions of wind vectors at future times, are provided as inputs to a machine learning model, such as a gradient-boosted decision trees algorithm, along with the surface roughness metrics and other relevant features. A continuous representation of predicted wind vectors over the ground region generated based on outputs received from the model is utilized to make operational decisions regarding an aerial vehicle.
G05D 1/606 - Compensation ou utilisation des conditions ambiantes externes, p. ex. du vent ou des courants d’eau
G01P 5/00 - Mesure de la vitesse des fluides, p. ex. d'un courant atmosphériqueMesure de la vitesse de corps, p. ex. navires, aéronefs, par rapport à des fluides
A solution builder service allows builder clients to build modules and solution templates for application deployment and also allows consumer clients to select solution templates for deployment of an application. The service may receive, from a builder, software code to be included in an application module and generate a module based on the code. After validation, the service adds the module to a catalog. The service may receive, from the builder, a solution template that indicates application module(s) of the catalog that are to be used for deployment of an application (e.g., solution). After validation, the service adds the template to the catalog. The template becomes available for selection and configuration by a consumer, which can be used to deploy an application to the consumer's resources.
A direct memory access (DMA) engine may receive a first indication that memory descriptors for a DMA operation are ready to be fetched from a memory. The DMA engine may prefetch the memory descriptors from the memory without waiting for memory locations specified in the memory descriptors to be ready for access. The DMA engine may receive a second indication that the memory locations specified in the memory descriptors are ready to be accessed. The DMA engine may execute the DMA operation based on the memory descriptors upon receiving the second indication.
G06F 13/28 - Gestion de demandes d'interconnexion ou de transfert pour l'accès au bus d'entrée/sortie utilisant le transfert par rafale, p. ex. acces direct à la mémoire, vol de cycle
Techniques for performing collective compute operations are described. A collective compute operation can be performed in a logical ring of processing ranks formed by a set of rank groups that each contain a number of processing ranks including a primary rank and one or more secondary ranks. At each rank group, a primary rank receives an incoming data slice via an intranode interconnect from a previous primary rank at multiple hops away on the logical ring. A data transfer is performed between the primary rank and each secondary rank of the rank group. An outgoing data slice is then transferred from the primary rank of the rank group to the next primary rank at multiple hops away on the logical ring.
G06F 15/173 - Communication entre processeurs utilisant un réseau d'interconnexion, p. ex. matriciel, de réarrangement, pyramidal, en étoile ou ramifié
G06F 13/28 - Gestion de demandes d'interconnexion ou de transfert pour l'accès au bus d'entrée/sortie utilisant le transfert par rafale, p. ex. acces direct à la mémoire, vol de cycle
Techniques for finetuning a large language model (LLM) are described. The finetuned LLM is generated by sequentially tuning a trained LLM using: (i) pairs of questions and corresponding context received from at least one source different from the trained LLM; and (ii) the questions without the context. The finetuned LLM may be used to generate an answer to a question, and user feedback and context may be used to update a trained machine learning (ML) configured to determine context for input to the finetuned LLM at inference. The updated trained ML model may thereafter process a question to determine context usable to answer the question; the question and the context may be processed by the finetuned LLM to determine an answer to the question; and the question, context, and answer may be used to further finetune the LLM.
Described are systems and methods that guide an LLM in determining when to initiate a request to a user for clarification and when to cause execution of a determined directive based on an utterance. At each pass through the LLM it may be determined whether the probability score and/or confidence score of a determined token exceeds respective thresholds. Based on those determinations, the system may decide whether to proceed with the determined tokens or to request clarification.
A technique for packing matrix multiplications for concurrent execution in an integrated circuit device may include obtaining a description of a neural network model, and generating an intermediate representation of the neural network model. Matrix multiplication instructions in the intermediate representation of the neural network model can then be vectorized for concurrent execution on an integrated circuit device, and machine instructions can be generated for the integrated circuit device based on the vectorized matrix multiplication instructions.
A processing engine array is provided with an interconnect mode of operation to use the array as an interconnect to move data elements to different locations in memory such as to perform a matrix transpose operation. In this interconnect mode of operation, although computations are still being performed in the array, the computations are not carried out to modify or change the values of the data elements, but are instead carried out to rearrange the data elements in memory. As such, the computations carried out in the interconnect mode of operation can deviate from the expected behavior of floating-point calculations. A mode selection signal can be used to provide the proper outputs of the processing elements of the array depending on the mode of operation.
G06F 7/78 - Dispositions pour le réagencement, la permutation ou la sélection de données selon des règles prédéterminées, indépendamment du contenu des données pour changer l'ordre du flux des données, p. ex. transposition matricielle ou tampons du type pile d'assiettes [LIFO]Gestion des occurrences du dépassement de la capacité du système ou de sa sous-alimentation à cet effet
G06F 12/0802 - Adressage d’un niveau de mémoire dans lequel l’accès aux données ou aux blocs de données désirés nécessite des moyens d’adressage associatif, p. ex. mémoires cache
G06N 3/063 - Réalisation physique, c.-à-d. mise en œuvre matérielle de réseaux neuronaux, de neurones ou de parties de neurone utilisant des moyens électroniques
