Abstract: A computer-implemented method for grouping devices in a massive multiple-input and multiple-output (MIMO)-based cellular network, in accordance with one embodiment, includes determining movement states of end devices in a cell of the massive MIMO-based cellular network, estimating payload requirements of the end devices, and grouping the end devices in a group based on the determined movement states and the estimated payload requirements.

Abstract: Computer-implemented methods for autonomous identification of rouge devices in a communications network are provided. Aspects include collecting connection trace data including connection characteristics for each of a plurality of mobile devices in communication with a communications network and aggregating the connection trace data. Aspects also include determining performance characteristics for each of a plurality of groups of the plurality of mobile devices, wherein each of the plurality of groups corresponds to mobile devices having a type approval code and comparing the performance characteristics for each of the plurality of groups. Based on a determination that the performance characteristics of one of the plurality of groups deviate from the performance characteristics of a remaining set of the plurality of groups by more than a first threshold amount, aspects include designating the type approval code associated with the one of the plurality of groups as a rogue type approval code.

Abstract: A disclosed framework enables virtually any aggressive Recovery Point Objective (RPO) to be enforced for a production database, without limitation on type of database management system (DBMS) or size of the production database. The framework overcomes technological barriers of the prior art, such as bottlenecks presented by a storage manager computer that traditionally managed database backups. To avoid the bottlenecks, an illustrative data storage management system does not engage the storage manager computer in storage operations that could impact the database RPO. Certain components are added and enhanced to autonomously capture database transaction logs, which are stored securely and persistently away from the DBMS. In a separate and less frequent operation, and with involvement of the storage manager computer, the illustrative system generates secondary copies in their final form. Meanwhile, temporary backup files can be used in restore operations without first being converted into secondary copies.

Abstract: Disclosed methods and systems leverage resources in a storage management system to restore a selected backup to a production site. The backup is partitioned into blocks with associated signatures. The production site may have blocks that have not changed from when the backup occurred, so those blocks do not need to be restored. Block signatures from the production site are compared with block signatures from the incremental backup to identify blocks that need to be restored. Efficiency may be achieved by synchronizing the replacement blocks from more easily accessible location where available before synchronizing from less accessible locations. In some embodiments, a user may specify the location of the site with the replacement blocks.

Abstract: The disclosed system and method focus on automatically generating questions from input of written text and/or audio transcripts (e.g., learning materials) to aid in teaching people through testing their knowledge about information they have previously been presented with. These questions may be presented to an end user via a conversational system (e.g., virtual agent or chatbot). The user can iterate through each question, provide feedback for the question, attempt to answer the question, and/or get an answer score for each answer. The disclosed system and method can generate questions tailored to a particular subject by using teaching materials as input. The disclosed system and method can further curate the questions based on various conditions to ensure that the questions are automatically selected and arranged in an order that best suits the subject taught and the learner answering the questions.

Abstract: Techniques disclosed herein are well suited to restoring deduplicated backup data from cloud-based storage and from multi-node replicated files systems, and they also improve performance in more traditional data storage technologies. Pre-restore steps include analysis of deduplication indexes to identify data segments that are stored consecutively on storage media. Reading data in aggregate runs of consecutively stored data segments reduces interactions with storage media that hosts the deduplicated data and speeds up retrieval. Parallel reads from multiple storage devices in multi-node replicated file systems also speed up retrieval. An illustrative enhanced media agent pre-fetches data (stored in deduplicated form) in anticipation of read requests that are expected in the restore operation.

Abstract: A recommender system can include a defender computing device that is configured to obtain customer interaction data characterizing customer interactions with an ecommerce marketplace. The defender computing device can also be configured to determine an item recommendation based on the customer interaction data using a trained differentially private recommendation model and send the item recommendation to the customer. The trained differentially private recommendation model is more likely to determine the same item recommendation after poisoned data is injected into the customer interaction data than a recommendation model that is not privately trained.

Abstract: This application relates to apparatus and methods for automatically detecting attacks to advertisement systems. In some examples, a computing device trains a machine learning process based on a training dataset. The training dataset may be an identified portion of a website session dataset that includes a lower percentage of malicious data caused by attacks than other portions, or may include no malicious data. Once trained, the computing device generates features from a website session dataset for a customer, and applies the trained machine learning process to the generated features to detect malicious data within the website session dataset for the customer. Further, the computing device may filter the website session data to remove the detected malicious data, and may store the filtered website session data within a data repository. The computing device may provide the filtered website session data to a recommendation system to generate item recommendations for the customer.

Abstract: Examples described herein include systems and methods for providing push notifications to a third-party application executing on a client device. An example can include encrypting user credentials, generating a callback Uniform Resource Locator (“URL”) with at least a portion of the encrypted credentials embedded into the URL, and requesting notifications from an email service to be provided at the callback URL. Upon receiving a notification at the callback URL, a system component can decrypt the credentials within the URL using a private key and log into the email account using those decrypted credentials. The system component can then generate a push notification based on any changes found in the email account and cause the notification to be delivered to the third-party application on the client device.

Abstract: An apparatus for winding coils onto a spherical body includes a frame, a feeder spool, a first hemispherical bobbin, a second hemispherical bobbin, a first spring-loaded pin, a second spring-loaded pin, and a motor arrangement. The feeder spool is rotationally mounted on the frame, has first and second wires wound thereon, and is configured to rotate about a first rotational axis. The first and second hemispherical bobbins are rotationally mounted on the frame and configured to rotate about a second rotational axis that is parallel to the first rotational axis. The bobbins are spaced apart from each other to define a wire-feeder gap through which the first and second wires may be fed. The motor arrangement is coupled to the first and second hemispherical bobbins and is configured to cause the first and second hemispherical bobbins to rotate in opposite directions about the second rotational axis.

Abstract: The disclosed system and method focus on automatically generating questions from input of written text and/or audio transcripts (e.g., learning materials) to aid in teaching people through testing their knowledge about information they have previously been presented with. These questions may be presented to an end user via a conversational system (e.g., virtual agent or chatbot). The user can iterate through each question, provide feedback for the question, attempt to answer the question, and/or get an answer score for each answer. The disclosed system and method can generate questions tailored to a particular subject by using teaching materials as input. The disclosed system and method can further curate the questions based on various conditions to ensure that the questions are automatically selected and arranged in an order that best suits the subject taught and the learner answering the questions.

Abstract: Systems and methods of real-time modeling pipeline inferencing are disclosed. At least one model configured to calculate at least one metric from one or more features is deployed. A model inferencing pipeline configured to extract the one or more features from a customer-specific data pipeline is implemented for the at least one mode. The model inferencing pipeline is generated using a training data set extracted from a cross-customer data pipeline. The at least one metric is calculated using the one or more features extracted from the customer-specific data pipeline.

Abstract: Techniques disclosed herein are used for restoring deduplicated backup data from cloud-based storage and from multi-node replicated files systems, and they also may be used in more traditional data storage technologies. Pre-restore steps include analysis of deduplication indexes to identify data segments that are stored consecutively on storage media. Reading data in aggregate runs of consecutively stored data segments reduces interactions with storage media that hosts the deduplicated data and speeds up retrieval. Parallel reads from multiple storage devices in multi-node replicated file systems also speed up retrieval. An illustrative enhanced media agent pre-fetches data (stored in deduplicated form) in anticipation of read requests that are expected in the restore operation.

Abstract: An information management system provides a data deduplication system that uses a primary table, a deduplication chunk table, and a chunk integrity table to ensure that a referenced deduplicated data block is only verified once during the data verification of a backup or other replication operation. The data deduplication system may reduce the computational and storage overhead associated with traditional data verification processes. The primary table, the deduplication chunk table, and the chunk integrity table, all of which may be stored in a deduplication database, can also ensure synchronization between the deduplication database and secondary storage devices.

Abstract: A privacy system includes a computing device configured to obtain user transactional data characterizing at least one transaction of a user on an ecommerce marketplace and to determine a privacy vulnerability score of the user by comparing the transactional data to a user vulnerability distribution. The computing device is also configured to send the privacy vulnerability score to a personalization engine.

Abstract: A recommender system can include a defender computing device that is configured to obtain customer interaction data characterizing customer interactions with an ecommerce marketplace. The defender computing device can also be configured to determine an item recommendation based on the customer interaction data using a trained differentially private recommendation model and send the item recommendation to the customer. The trained differentially private recommendation model is more likely to determine the same item recommendation after poisoned data is injected into the customer interaction data than a recommendation model that is not privately trained.

Abstract: This application relates to apparatus and methods for automatically detecting attacks to advertisement systems. In some examples, a computing device trains a machine learning process based on a training dataset. The training dataset may be an identified portion of a website session dataset that includes a lower percentage of malicious data caused by attacks than other portions, or may include no malicious data. Once trained, the computing device generates features from a website session dataset for a customer, and applies the trained machine learning process to the generated features to detect malicious data within the website session dataset for the customer. Further, the computing device may filter the website session data to remove the detected malicious data, and may store the filtered website session data within a data repository. The computing device may provide the filtered website session data to a recommendation system to generate item recommendations for the customer.

Abstract: A spherical brushless direct current (BLDC) machine includes a first stator, a second stator, and a spherical rotor. The first stator is symmetrically disposed about a first axis and includes a first multi-pole stator core having a first multi-phase winding wound thereon. The second stator is symmetrically disposed about a second axis and includes a second multi-pole stator core having a second multi-phase winding wound thereon. The second stator core is coupled to the first stator core, and the second axis intersects the first axis. The spherical rotor is disposed adjacent to, and is moveable relative to, the first and second stators. The spherical rotor includes a plurality of magnets that emanate a magnetic field, and each magnet has at least one of its magnetic poles facing the first and second stators.

Abstract: A high availability distributed, deduplicated storage system according to certain embodiments is arranged to include multiple deduplication database media agents. The deduplication database media agents store signatures of data blocks stored in secondary storage. In addition, the deduplication database media agents are configured as failover deduplication database media agents in the event that one of the deduplication database media agents becomes unavailable.

Abstract: Disclosed are various examples for threat detection and security for edge devices in communication with Internet-of-Things (IoT) devices. In one example, a profile is associated with a virtual machine of a gateway device. The profile includes an expected behavior for the virtual machine. The virtual machine is executed by a hypervisor of the gateway device. An actual behavior for the virtual machine is determined. A remedial action is performed. The remedial action is based on an anomaly between the expected behavior and the actual behavior.