Abstract: Aspects of the present invention relate to an active grille shutter vane for an active grille shutter assembly. The vane comprises: a leading edge and a trailing edge whereby, in use, the vane is configured such that airflow passes around the vane from the leading edge to the trailing edge when the active grille shutter assembly is in an open configuration; and a primary surface extending between the leading edge and the trailing edge. The leading edge, primary surface and trailing edge provide a continuous cambered attachment surface for the airflow passing over the primary surface to encourage attachment of the airflow to the vane.

Abstract: Systems and method for visualizing data for an automated question answering system are disclosed. The method includes identifying a first answer associated with a first query; identifying a second answer associated with a second query; determining a criterion of the first answer with respect to the second answer; and visually representing the first answer and the second answer based on the criterion.

Abstract: Techniques are disclosed for detecting adversarial attacks. A machine learning (ML) system processes the input into and output of a ML model using an adversarial detection module that does not include a direct external interface. The adversarial detection module includes a detection model that generates a score indicative of whether the input is adversarial using, e.g., a neural fingerprinting technique or a comparison of features extracted by a surrogate ML model to an expected feature distribution for the output of the ML model. In turn, the adversarial score is compared to a predefined threshold for raising an adversarial flag. Appropriate remedial measures, such as notifying a user, may be taken when the adversarial score satisfies the threshold and raises the adversarial flag.

Abstract: Various examples are directed to computer-implemented systems and methods for providing a triage tool for investment advising. A method includes displaying, by a processor of a computer, an image on a graphical user interface (GUI) of a device of a user. The image includes information related to financial account management for the user. Input options are provided to the user on the GUI including at least two directional swipe input options to obtain the user's positive or negative response to the image. The processor receives the user's response to the input options, and determines a profile of the user based on the user's response to the input options. The processor provides directed financial investment advice and tailored financial products, based on the profile of the user, to increase applicability and effectiveness of wealth management for the user.

Abstract: Training a user-specific perturbation generator for an audio feature detection model includes receiving one or more positive audio samples of a user, each of the one or more positive audio samples including an audio feature; receiving one or more negative audio samples of the user, each of the one or more negative audio samples sharing an acoustic similarity with at least one of the one or more positive audio samples; and adversarially training a user-specific perturbation generator model to generate a user-specific perturbation, the training based on the one or more positive audio samples and the one or more negative audio samples. Perturbing audio samples of the user with the user-specific perturbation can cause an audio feature detection model to recognize the audio feature in audio samples that include the audio feature and/or to refrain from recognizing the audio feature in audio samples that do not include the audio feature.

Abstract: The current disclosure relates to systems and methods for wakeword or keyword detection in Virtual Personal Assistants (VPAs). In particular, systems and methods are provided for wakeword detection using deep neural networks including a parametric pooling layer, wherein the parametric pooling layer includes trainable parameters, enabling the layer to learn to distinguish between informative feature vectors and non-informative/noisy feature vectors extracted from a variable length acoustic signal. In one example, a parametric pooling layer may aggregate a variable length feature map, comprising a plurality of feature vectors extracted from an acoustic signal, into an embedding vector of pre-determined length, by weighting each of the plurality of feature vectors based on one or more learned parameters in a parametric pooling layer, and aggregating the plurality of weighted feature vectors into the embedding vector.

Abstract: Techniques are disclosed for detecting adversarial attacks. A machine learning (ML) system processes the input into and output of a ML model using an adversarial detection module that does not include a direct external interface. The adversarial detection module includes a detection model that generates a score indicative of whether the input is adversarial using, e.g., a neural fingerprinting technique or a comparison of features extracted by a surrogate ML model to an expected feature distribution for the output of the ML model. In turn, the adversarial score is compared to a predefined threshold for raising an adversarial flag. Appropriate remedial measures, such as notifying a user, may be taken when the adversarial score satisfies the threshold and raises the adversarial flag.

Abstract: Various examples are directed to computer-implemented systems and methods for providing a triage tool for investment advising. A method includes displaying by a processor of a computer, an image on a graphical user interface (GUI) of a device of a user. The image includes information related to financial account management for the user. Input options are provided to the user on the GUI including at least two directional swipe input options to obtain the user's positive or negative response to the image. The processor receives the user's response to the input options, and determines a profile of the user based on the user's response to the input options. The processor provides directed financial investment advice and tailored financial products, based on the profile of the user, to increase applicability and effectiveness of wealth management for the user.

Abstract: A method for inspecting an object by ultrasound for detecting a wall thickness or defects of the object, wherein at least one ultrasonic pulse is transmitted into the object on a first position on an object's surface, the ultrasonic pulse is received on a second position on an object's surface possibly by propagating directly towards the second position along the surface or possibly as a result of reflection and/or diffractions of the pulse so that more than one pulse being received at different time and wherein a data signal is generated representing the received pulses and the associated moments in time wherein these pulses are received wherein the step is repeated for other positions and wherein the data signals are processed for generating processed signals to obtain a To FD image, wherein the processing for obtaining the processed data signals comprises at least three processing steps.

Abstract: A method for inspecting an object by means of ultrasound, for example for detecting a wall thickness of the object or for detecting defects in the object, wherein in a step a. at least one ultrasonic pulse is transmitted into the object on a first position of a surface of the object, the ultrasonic pulse is received on a second position of the surface of the object possibly by propagating directly towards the second position along the surface of the object or possibly as a result of reflection and/or diffractions of the pulse in the object so that the transmitted pulse may result in more than one pulse being received at different moments in time and wherein a data signal is generated representing the received pulses and the associated moments in time wherein these pulses are received wherein step a. is repeated for other first positions and/or second positions for generating a first set of data signals and wherein the first set of data signals obtained in the plurality of steps a.

Abstract: A method for ultrasonic inspecting a substantially circular circumferential weld which connects an inner pipe and an outer pipe which are aligned coaxial relative to each other, the method comprising the steps of: a) positioning at least one ultrasonic probe inside the inner pipe; b) moving the at least one probe in tangential direction relative to the inner pipe a while emitting at least one ultrasonic beam towards the weld and receiving ultrasonic signals by means of the at least one probe.