Abstract: The disclosed embodiments disclose techniques for performing universal post-training backdoor detection and mitigation for classifiers. Mitigation of overfitting for a trained classifier begins with receiving the trained classifier and a clean dataset that spans a plurality of classes for the trained classifier. A set of input patterns are used to calculate classification margins for the trained classifier, and maximum classification margins are calculated for one or more classes of the trained classifier. Overfitting can then be mitigated by reducing one or more of these calculated maximum classification margins while maintaining the accuracy of the trained classifier for the clean dataset. In some embodiments, a backdoor detector may also detect target classes for a putative backdoor in the trained classifier upon detecting that the corresponding maximum classification margins for those target classes are anomalously high compared to the maximum classification margins of other classes.

Abstract: Systems and methods for a radio communications device supporting LTE cellular communications. A device includes a data interface configured to exchange modem control commands and audio data with an LTE cellular modem and a telephone interface configured to provide a two-wire analog telephone interface for connection to a telephone device. A controller control signals provided via the telephone interface to manage voice call operations with the telephone device; exchange digital voice data corresponding to voice signals exchanged with the telephone device via the telephone interface; exchange control information via the data interface to establish and manage a voice over LTE cellular call; and exchange the digital voice data via the voice over LTE cellular call via the LTE cellular modem.

Abstract: A motor controller is described that is coupled to a drive motor and a battery pack of a vehicle. The motor controller is configured to determine a maximum discharge current of the battery pack and a rotational velocity of the drive motor. Based on the determined rotational velocity of the drive motor, the motor controller is configured to identify a curve that defines a relationship between the maximum discharge current of the battery pack and a drive current limit of the motor controller. Based on the identified curve and the determined maximum discharge current of the battery pack, the motor controller is configured to determine the drive current limit of the motor controller. The motor controller is further configured to convert a discharge current from the battery pack to a drive current subject to the determined drive current limit and supply the drive current to the drive motor.

Abstract: In examples provided herein, upon receiving notification of a computational task requested by a package to provide an experience to a user, a remote node management engine identifies computing nodes for performing the computational task and determining available processing resources for each computing node, where a computing node resides at networked wearable devices associated with the user. The remote node management engine further selects one of the computing nodes as a primary controller to distribute portions of the computational task to one or more of the other computing nodes and receive results from performance of the portions of the computational task by the other computing nodes, and provides to the selected computing node information about available processing resources at each computing node.

Abstract: This disclosure relates to novel mjresynthesis kits and methods, in particular for use in pairwise sequencing.

Abstract: The present invention provides compositions comprising peptide-coupled biodegradable poly(lactide-co-glycolide) (PLG) particles. In particular, PLG particles are surface-functionalized to allow for coupling of peptide molecules to the surface of the particles (e.g., for use in eliciting induction of immunological tolerance).

Abstract: The present invention relates to sulfonylureas and sulfonylthioureas comprising a monocyclic imidazolyl group of formula (I): wherein Q is selected from O or S, R1 is an imidazolyl group, wherein the imidazolyl group is unsubstituted or substituted with one or more monovalent substituents, and R2 is a cyclic group substituted at the ?-position, wherein R2 may optionally be further substituted. The present invention further relates to salts, solvates and prodrugs of such compounds, to pharmaceutical compositions comprising such compounds, and to the use of such compounds in the treatment and prevention of medical disorders and diseases, most especially by the inhibition of NLRP3.

Abstract: A hitch mounted tree carrier is configured to be secured to a motor vehicle and permits a user to transport a tree from a first location to a second location. The hitch mounted tree carrier includes an elongated member configured to be removably received by a motor vehicle hitch and a receptacle having a front portion and a rear portion, wherein the front portion is affixed to the elongated member. The hitch mounted tree carrier also includes: (i) a wedge positioned in the receptacle and configured to apply a first force on a first extent of a tree trunk, when the tree is in the process of being inserted into the receptacle, and (ii) a securement assembly extending into the receptacle and configured to apply a second force on a second extent of the tree trunk, when the tree is in an engaged state.

Abstract: Methods and systems are provided for determining the material composition of subterranean metal objects. By propagating an electromagnetic field through the ground and impinging upon the metal object, eddy currents form that generate a secondary magnetic field that can be measured. The measured field can be compared to existing, model data to determine the material composition of the metal object.

Abstract: Systems and methods for configuring operation of an engine are described herein. A computer-readable label associate with the engine is read by a mobile device to obtain label information having at least one trim value for the engine encoded therein. The at least one trim value is extracted from the label information on the mobile device. The at least one trim value is wirelessly transmitted from the mobile device to a data transmission unit of the engine. The data transmission unit is configured for instructing an electronic engine controller to trim the engine with the at least one trim value during operation of the engine.

Abstract: Execution of an executable is initiated in an operating environment. Thereafter, characteristics associated with the execution of the executable are continually monitored. It is later determined that the executable has one or more characteristics associated with ransomware. In response to such determination, at least one simulacra is injected into the operating environment to attempt to trigger an anti-detonation protection feature by the executable file. Thereafter, one or more ransomware countermeasures can be initiated based on behavior of the executable after injection of the at least one simulacra. Related apparatus, systems, techniques and articles are also described.

Abstract: Data is received that comprises or characterizes an executable and dynamic linked library (DLL). Features are then extracted from the executable and DLL. The extracted features are input into at least one machine learning model to generate a suspiciousness score. The machine learning model can be trained to determine whether the executable file comprises ransomware. An execution chain of trust score for the executable and DLL can later be determined based on the extracted features and the suspiciousness score. This execution chain of trust score for the executable and DLL characterizes one or more associated parent processes. This suspiciousness score and the execution chain of trust score can be used to determine whether or not to initiate one or more ransomware countermeasures. Related apparatus, systems, techniques and articles are also described.

Abstract: An electronic system for a firearm includes a power source, one or more electrical conductors electrically connected to receive power from the power source, and a plurality of electronic devices. Each electronic device has an electrical input configured to receive power from the one or more electrical conductors to power the electronic device. A communication device is configured for data communication across the one or more electrical conductors. In some embodiments data can be communicated at rates greater than 10 Mbps. Some embodiments utilize multi-frequency encoding, such as orthogonal frequency-division multiplexing. Some embodiments include a data communication device with a lower data communication rate and another data communication device with a higher data communication rate, and can selectively communicate through either data communication device.

Abstract: This patent concerns novel technology for detecting backdoors in neural network, particularly deep neural network (DNN) classification or prediction/regression models. The backdoors are planted by suitably poisoning the training dataset, i.e., a data-poisoning attack. Once added to an input sample from a source class of the attack, the backdoor pattern causes the decision of the neural network to change to the attacker's target class in the case of classification, or causes the output of the network to significantly change in the case of prediction or regression. The backdoors under consideration are small in norm so as to be imperceptible to a human or otherwise innocuous/evasive, but this does not limit their location, support or manner of incorporation. There may not be components (edges, nodes) of the DNN which are specifically dedicated to achieving the backdoor function. Moreover, the training dataset used to learn the classifier or predictor/regressor may not be available.

Abstract: There is provided a data access method and system for an aircraft. Password data is obtained for use in providing access to aviation data collected by at least one data acquisition device from one or more locations in the aircraft. The password data is caused to be rendered on at least one display provided in at least one secured location of the aircraft. Subsequent to the password data being rendered on the at least one display, input data is received from at least one user device requesting access to the aviation data. The input data is compared to the password data and, when the input data matches the password data, a connection is established between the at least one user device and the at least one data acquisition device for providing the at least one user device access to the aviation data.

Abstract: Systems and methods for configuring operation of an engine are described herein. A computer-readable label associate with the engine is read by a mobile device to obtain label information having at least one trim value for the engine encoded therein. The at least one trim value is extracted from the label information on the mobile device. The at least one trim value is wirelessly transmitted from the mobile device to a data transmission unit of the engine. The data transmission unit is configured for instructing an electronic engine controller to trim the engine with the at least one trim value during operation of the engine.

Abstract: This patent concerns novel technology for detecting backdoors of neural network, particularly deep neural network (DNN), classifiers. The backdoors are planted by suitably poisoning the training dataset, i.e., a data-poisoning attack. Once added to input samples from a source class (or source classes), the backdoor pattern causes the decision of the neural network to change to a target class. The backdoors under consideration are small in norm so as to be imperceptible to a human, but this does not limit their location, support or manner of incorporation. There may not be components (edges, nodes) of the DNN which are dedicated to achieving the backdoor function. Moreover, the training dataset used to learn the classifier may not be available.

Abstract: An electronic system for a firearm includes a power source, one or more electrical conductors electrically connected to receive power from the power source, and a plurality of electronic devices. Each electronic device has an electrical input configured to receive power from the one or more electrical conductors to power the electronic device. A communication device is configured for data communication across the one or more electrical conductors. In some embodiments data can be communicated at rates greater than 10 Mbps. Some embodiments utilize multi-frequency encoding, such as orthogonal frequency-division multiplexing. Some embodiments include a data communication device with a lower data communication rate and another data communication device with a higher data communication rate, and can selectively communicate through either data communication device.

Abstract: A packaging card for a tool comprising a fixed member and a movable member that is selectively movable relative to the fixed member includes a tool engaging region coupling the packaging card to the movable member such that movement of the movable member relative to the fixed member moves the card relative to the fixed member. The card also includes a deformation region that tears or deforms upon application of a separation force sufficient to remove the packaging card from the tool. Another packaging card for a tool having a housing with front and side apertures includes a front flange receivable in the front aperture of the tool, a side flange insertable into the side aperture, and a hook catchable on an interior of the housing to deter removal of the packaging card from the tool until sufficient force is applied to overcome a holding power of the hook.

Abstract: This patent concerns novel technology for detecting backdoors of neural network, particularly deep neural network (DNN), classifiers. The backdoors are planted by suitably poisoning the training dataset, i.e., a data-poisoning attack. Once added to input samples from a source class (or source classes), the backdoor pattern causes the decision of the neural network to change to a target class. The backdoors under consideration are small in norm so as to be imperceptible to a human, but this does not limit their location, support or manner of incorporation. There may not be components (edges, nodes) of the DNN which are dedicated to achieving the backdoor function. Moreover, the training dataset used to learn the classifier may not be available.