Abstract: In one embodiment, a device forms a neural network envelope cell that comprises a plurality of convolution-based filters in series or parallel. The device constructs a convolutional neural network by stacking copies of the envelope cell in series. The device trains, using a training dataset of images, the convolutional neural network to perform image classification by iteratively collecting variance metrics for each filter in each envelope cell, pruning filters with low variance metrics from the convolutional neural network, and appending a new copy of the envelope cell into the convolutional neural network.

Abstract: Systems and methods for distributing over the air (OTA) content are provided. In one example, an antenna system can include a network access point. The antenna system can include an OTA antenna system configured to receive an OTA signal associated with media content. The OTA signal can be a VHF signal or a UHF signal. The OTA antenna system can be coupled to the network access point via a universal serial bus (USB) connection. The access point can be configured to perform operations, including receiving a signal associated with the media content from the OTA antenna system via the universal serial bus connection; and communicating media content via a network communication link to one or more client devices. The OTA antenna system can include a modal antenna configurable in a plurality of antenna modes. Each antenna mode can be associated with a distinct radiation pattern or polarization.

Abstract: Systems and methods for detecting injection exploits in a networked computing environment are disclosed. In one embodiment, a method for detection includes monitoring web applications that are executing and detecting when an execution function is received over a network and invoked, where an execution function is a function that accepts external free-form data values, detecting malicious code by generating a model of legitimate behavior subsequent to invocation of the execution function, comparing actual behavior to the model of legitimate behavior, and generating an alert when the actual behavior deviates from the model of legitimate behavior and validating whether the deviation of the actual behavior is due to one or more functions that accept external input.

Abstract: The disclosure concerns an antenna subsystem that can be used in various repeater systems to optimize gain of the repeater by increasing isolation between donor and server antennas, wherein at least one of the donor and server antennas is an active multi-mode antenna.

Abstract: A system and method for generating an alert regarding a potential attack is described. The method involves receiving data associated with previously analyzed or known malware attacks by a first network device. Additionally, the first network device receives an attack alert associated with an object analyzed and identified as suspicious by a second network device. The attack alert includes information associated with the suspicious object. For alert generation, at least a portion of the information of the attack alert is provided to a system configured to at least (i) extract feature(s) from the attack alert, (ii) determine similarities between the extracted features and features associated with the previously analyzed or known malware attacks to determine a result, (iv) compute an attack value based on the result and at least a portion of the extracted features including time-dependent and/or independent features, and (v) generate an alert based on the attack value.

Abstract: The disclosure pertains to techniques for image processing. One such technique comprises a method for image selection, comprising: obtaining a sequence of images, detecting a first face in one or more images of the sequence of images, determining a first location for the detected first face in each of the images having the detected first face, generating a heat map based on the first location of the detected first face in each of the images of the sequences of images, determining a face quality score for the detected first face for each of the one or more images having the detected first face, determining a peak face quality score for the detected first face based in part on the face quality score and the generated heat map, and selecting a first image of the sequence of images, corresponding with the peak face quality score for the detected first face.

Abstract: A distributed deep learning network may prevent an attacker from reconstructing raw data from activation outputs of an intermediate layer of the network. To achieve this, the loss function of the network may tend to reduce distance correlation between raw data and the activation outputs. For instance, the loss function may be the sum of two terms, where the first term is weighted distance correlation between raw data and activation outputs of a split layer of the network, and the second term is weighted categorical cross entropy of actual labels and label predictions. Distance correlation with the entire raw data may be minimized. Alternatively, distance correlation with only with certain features of the raw data may be minimized, in order to ensure attribute-level privacy. In some cases, a client computer calculates decorrelated representations of raw data before sharing information about the data with external computers.

Abstract: The disclosure describes a tool that decreases the size of a software installation file by segregating and/or allowing customization of the original software installation file. The resulting install file permits multiple configurations to be created using the same install file by using custom configuration settings. The optimized software installation file permits the automated (or semi-automated) installation of software and/or software updates/upgrades on a remotely located electronic device. Moreover, aspects of the tool may also regulate compliance with possible reboot/restart requirements of the software/upgrade installation that may be useful for electronic devices which are remotely located.

Abstract: Techniques for implementing zone redundant computing services in distributed computing systems are described herein. In one embodiment, a computing device is configured to query a database in a first availability zone for a list of computing services corresponding to a user account and provided by servers in a second availability zone different than the first availability zone. The computing device can then selects one or more computing services provided by the servers in the second availability zone from the queried list of computing services and transmit an instruction to the servers in the second availability zone for performing the requested data operation.

Abstract: The present disclosure provides systems, methods and computer-readable media for optimizing the neural architecture search for the automated machine learning process. In one aspect, neural architecture search method including selecting a neural architecture for training as part of an automated machine learning process; collecting statistical parameters on individual nodes of the neural architecture during the training; determining, based on the statistical parameters, active nodes of the neural architecture to form a candidate neural architecture; and validating the candidate neural architecture to produce a trained neural architecture to be used in implemented an application or a service.

Abstract: A multi-mode antenna system include at least a first modal antenna and a second modal antenna. The first modal antenna is disposed on a ground plane of a circuit board and configurable in a plurality of different modes. The first modal antenna can include a driven element, at least one parasitic element and an active element configured to adjust a reactance of the at least one parasitic element. The multi-mode antenna system further includes a second modal antenna disposed on the ground plane and configurable in a plurality of different modes. The second modal antenna can include a driven element, at least one parasitic element, and an active element configured to adjust a reactance of the at least one parasitic element. The parasitic element of the second modal antenna is positioned such that adjusting the reactance of the parasitic element affects the radiation pattern associated with the first modal antenna.

Abstract: A neural network architecture search may be conducted by a controller to generate a neural network. The controller may perform the search by generating a directed acyclic graph across nodes in a search space, the nodes representing compute operations for a neural network. As the search is performed, the controller may retrieve resource availability information to modify the likelihood of a generated neural network architecture including previously unused nodes.

Abstract: An audio circuit includes an amplifier, a voltage sensor, a current sensor, and an excursion control circuit. The voltage sensor is coupled to an output of the amplifier. The current sensor is coupled to the output of the amplifier. The excursion control circuit is coupled to the amplifier, the voltage sensor, and the current sensor. The excursion control circuit includes back electro-magnetic force (EMF) measurement, a back-EMF model, and excursion protection. The back-EMF measurement is to measure back electro-magnetic force of a speaker based on voltage measurements received from the voltage sensor and current measurements received from the current sensor. The back-EMF model is updated based on measurements of the back-EMF and is converted to an excursion model. The excursion protection is to limit amplitude of audio signal provided to the amplifier based on the excursion model of the speaker and amplitude of an audio input signal.

Abstract: The disclosure concerns an antenna subsystem that can be used in various repeater systems to optimize gain of the repeater by increasing isolation between donor and server antennas, wherein at least one of the donor and server antennas is an active multi-mode antenna.

Abstract: Disclosed herein is a touch screen controller including a driver circuit applying a drive signal to a drive line of a capacitive touch sensing panel. The driver circuit is powered by an accurate supply voltage. A driver supply circuit receives an input supply voltage and outputs the accurate supply voltage. The driver supply circuit includes a clocked comparator comparing a divided version of the accurate supply voltage to a reference voltage and outputting a comparison signal based thereupon. A voltage control circuit (e.g. a charge pump circuit) generates the accurate supply voltage in response to the comparison signal. The clocked comparator and voltage control circuit are both clocked by a driver supply circuit clock.

Abstract: Methods, systems, and apparatuses manage rolling out of updates in a network-accessible server infrastructure which operates a plurality of instances of a supporting service. The supporting service is comprised by a plurality of service portions. The instances of the supporting service each include of the service portions. The instances of the supporting service are partitioned into a plurality of slices. Each instance is partitioned to include one or more of the slices, and each slice of an instance includes one or more of the service portions. A software update is deployed to the instances by applying the software update to the slices in a sequence such that the software update is applied to a same slice in parallel across the instances containing that same slice before being applied to a next slice, and waiting a wait time before applying the software domain to a next slice in the sequencing.

Abstract: A mobile computing device including a modal antenna is disclosed. The mobile computing device may include a radio frequency circuit and a modal antenna mechanically coupled to a portion of the mobile computing at a location that is remote from the radio frequency circuit. The modal antenna may include a driven element and a parasitic element positioned proximate to the driven element. The modal antenna may be operable in a plurality of different modes. Each mode may be associated with a different radiation pattern. The mobile computing device may include a transmission line coupling the radio frequency circuit to the modal antenna. The radio frequency circuit may be configured to transmit an RF signal over the transmission line to the modal antenna and configured to communicate a control signal to adjust the mode of the modal antenna over the transmission line.

Abstract: An integrated circuit package may be formed having at least one heat dissipation structure within the integrated circuit package itself. In one embodiment, the integrated circuit package may include a substrate; at least one integrated circuit device, wherein the at least one integrated circuit device is electrically attached to the substrate; a mold material on the substrate and adjacent to the at least one integrated circuit device; and at least one heat dissipation structure contacting the at least one integrated circuit, wherein the at least one heat dissipation structure is embedded either within the mold material or between the mold material and the substrate.

Abstract: A method of detecting malware in a specimen of computer content or network traffic is described. The method features conducting a first analysis on the specimen in accordance with a first plurality of analyses and an order of the first plurality of analyses. A second analysis is conducted on the specimen different than the first analysis type. Thereafter, further analyses on the specimen may be altered by modifying information associated with the first plurality of analyses or the order of the first plurality of analyses in response to feedback information based on results from at least the first analysis. The modified information changes a malware analysis of the specimen from being conducted in accordance with the first plurality of analyses to being conducted in accordance with a second plurality of analyses different in analysis type or in order of analyses than the first plurality of analyses.

Abstract: The disclosed technology relates to intent driven network management. A system is configured to maintain an inventory store comprising records for a set of network entities in a network, wherein each network entity in the set of network entities is associated with a record in the inventory store. The system receives a user intent statement comprising an action and a flow filter representing network data flows on which the action is to be applied and queries, based on the flow filter, the inventory store to identify a plurality of network entities in the set of network entities to which the user intent statement applies. The system generates a plurality of network policies that implement the user intent statement based on the plurality of network entities and the action and enforces the plurality network policies.