Abstract: A method can include providing a first neural network connected to a second neural network. The first neural network can represent B+1 and the second neural network can represent electrical properties. The method can include training the first neural network and the second neural network jointly. The method can include determining, from the trained first neural network and the trained second neural network B+1, a prediction of and electrical properties at one or more predetermined locations. The method can include outputting the prediction of B+1 and EP.

Abstract: A computer implemented method for certifying robustness of image classification in a neural network is provided. The method includes initializing a neural network model. The neural network model includes a problem space and a decision boundary. A processor receives a data set of images, image labels, and a perturbation schedule. Images are drawn from the data set in the problem space. A distance from the decision boundary is determined for the images in the problem space. A re-weighting value is applied to the images. A modified perturbation magnitude is applied to the images. A total loss function for the images in the problem space is determined using the re-weighting value. A confidence level of the classification of the images in the data set is evaluated for certifiable robustness.

Abstract: Obtain, using at least one hardware processor, data characterizing a physical system governed by a physical conservation law. Apply, using the at least one hardware processor, contrastive learning to the data to automatically capture system invariants of the physical system. Employ, using the at least one hardware processor, a neural projection layer to guarantee that a corresponding dynamic machine learning model preserves the captured system invariants. Optionally, predict performance of the physical system using the corresponding dynamic machine learning model.

Abstract: A system and method for measuring a hydration level in human tissue. The system includes a coaxial probe have a first end configured to be in contract with human tissue and configured to be emit and receive signals associated with a spectroscopic technique and having a second end adapted to be coupled to transmit and receive circuitry. The system further includes a patch or other means coupled to the coaxial probe and configured to be adhered to human tissue so as to provide a force upon at least a portion of the coaxial probe configured to be in contact with human tissue. In embodiments, the system may further include means coupled to receive signals from the coaxial probe, for determining an amount of liquid within a portion of human tissue in contact with the first end of the probe.

Abstract: One or more computer processors create a fully convolution network (FCN) comprising a plurality of 1×1 convolutions. The one or more computer processors append linear mapping layer (LM) to created FCN. The one or more computer processors capture a plurality of features utilizing multi-scale dilated convolutional kernels from the linear mapped FCN (LM-FCN). The one or more computer processors apply an average pool layer to the captured plurality of features along a temporal axis of a dilated convolutional kernel within the LM-FCN. The one or more computer processors predict a classification for subsequent time-series data utilizing the pooled plurality of features.

Abstract: Providing a trained reinforcement learning (RL) model by formulating a decision process problem for the RL model, defining at least one of a logarithmic loss function for the RL model and defining an initiation point for the RL model according to an optimized spectral norm of the RL model, training the system according to the logarithmic loss function or from the initiation point, and providing the trained RL model.

Abstract: A computer implemented method for certifying robustness of image classification in a neural network is provided. The method includes initializing a neural network model. The neural network model includes a problem space and a decision boundary. A processor receives a data set of images, image labels, and a perturbation schedule. Images are drawn from the data set in the problem space. A distance from the decision boundary is determined for the images in the problem space. A re-weighting value is applied to the images. A modified perturbation magnitude is applied to the images. A total loss function for the images in the problem space is determined using the re-weighting value. A confidence level of the classification of the images in the data set is evaluated for certifiable robustness.

Abstract: A certification method, system, and computer program product include certifying an adversarial robustness of a convolutional neural network by deriving an analytic solution for a neural network output using an efficient upper bound and an efficient lower bound on an activation function and applying the analytic solution in computing a certified robustness.

Abstract: A plurality of stimulations is transmitted to tissue or other material using one or more transmitters. The plurality of signals associated with the excited tissue and the transmitted stimulations are measured. The measured signals are processed to generate field-related quantities, such as B1+ and/or MR signal maps. Field-related quantities are generated also from simulation, by calculating the one or more incident fields from a simulator model of the one or more transmitters and assuming a given distribution of electrical properties in the tissue or other material. Field-related quantities generated from simulation and experimental procedures are compared to each other. The assumed electrical properties distribution is updated and the procedure is repeated iteratively until the difference between simulated and experimental field-related quantities is smaller than a threshold.

Abstract: A certification method, system, and computer program product include certifying an adversarial robustness of a convolutional neural network by deriving an analytic solution for a neural network output using an efficient upper bound and an efficient lower bound on an activation function and applying the analytic solution in computing a certified robustness.

Abstract: A system and method for measuring a hydration level in human tissue. The system includes a coaxial probe have a first end configured to be in contract with human tissue and configured to be emit and receive signals associated with a spectroscopic technique and having a second end adapted to be coupled to transmit and receive circuitry. The system further includes a patch or other means coupled to the coaxial probe and configured to be adhered to human tissue so as to provide a force upon at least a portion of the coaxial probe configured to be in contact with human tissue. In embodiments, the system may further include means coupled to receive signals from the coaxial probe, for determining an amount of liquid within a portion of human tissue in contact with the first end of the probe.

Abstract: A method of determining a decoupling matrix of a decoupling system for an array of coils of a parallel transmission magnetic resonance imaging (MRI) system includes obtaining impedance matrix data for the array of coils without the decoupling system, determining, based on the impedance matrix data for the array of coils, an objective function representative of deviation from a decoupled operating condition for the array of coils in which the array of coils are decoupled via the decoupling system, and defining, with a processor, a decoupling matrix representative of a set of impedances of the decoupling system with an iterative procedure that optimizes elements of the decoupling matrix to minimize the objective function and reach the decoupled operating condition.

Abstract: A magnetic resonance imaging (MRI) system includes a plurality of transmitters to generate a parallel transmission radio frequency (RF) pulse, an array of coils coupled to the plurality of transmitters to apply the parallel transmission RF pulse to a subject, and a decoupling system connected to the plurality of transmitters and the array of coils. The decoupling system includes a plurality of hybrid couplers, each hybrid coupler of the plurality of hybrid couplers being coupled to a respective pair of the plurality of transmitters and to a respective pair of the array of coils. The plurality of hybrid couplers are configured to diagonalize an impedance matrix of the plurality of coils.

Abstract: A plurality of stimulations is transmitted to tissue or other material using one or more transmitters. The plurality of signals associated with the excited tissue and the transmitted stimulations are measured. The measured signals are processed to generate field-related quantities, such as B1+ and/or MR signal maps. Field-related quantities are generated also from simulation, by calculating the one or more incident fields from a simulator model of the one or more transmitters and assuming a given distribution of electrical properties in the tissue or other material. Field-related quantities generated from simulation and experimental procedures are compared to each other. The assumed electrical properties distribution is updated and the procedure is repeated iteratively until the difference between simulated and experimental field-related quantities is smaller than a threshold.

Abstract: A magnetic resonance imaging (MRI) system includes a plurality of transmitters to generate a parallel transmission radio frequency (RF) pulse, an array of coils coupled to the plurality of transmitters to apply the parallel transmission RF pulse to a subject, and a decoupling system connected to the plurality of transmitters and the array of coils. The decoupling system includes a plurality of hybrid couplers, each hybrid coupler of the plurality of hybrid couplers being coupled to a respective pair of the plurality of transmitters and to a respective pair of the array of coils. The plurality of hybrid couplers are configured to diagonalize an impedance matrix of the plurality of coils.

Abstract: A method of determining a decoupling matrix of a decoupling system for an array of coils of a parallel transmission magnetic resonance imaging (MRI) system includes obtaining impedance matrix data for the array of coils without the decoupling system, determining, based on the impedance matrix data for the array of coils, an objective function representative of deviation from a decoupled operating condition for the array of coils in which the array of coils are decoupled via the decoupling system, and defining, with a processor, a decoupling matrix representative of a set of impedances of the decoupling system with an iterative procedure that optimizes elements of the decoupling matrix to minimize the objective function and reach the decoupled operating condition.

Abstract: A reduced order model of a distributed time invariant system is produced by projecting system matrices onto smaller matrices, interpolating the matrices and placing into a state-space system. The system matrices are an internal representation of the distributed time invariant system which comprises a description of the system to be modeled, mainly, for example, its inputs and outputs. The method is applied to distributed systems and guarantees accuracy in complicated systems and produces well-behaved models appropriate for use in simulators and simulations.