Abstract: Present disclosure discloses an image processing system and method for manipulating two-dimensional (2D) images of three-dimensional (3D) objects of a predetermined class (e.g., human faces). A 2D input image of a 3D object of the predetermined class is manipulated by manipulating physical properties of the 3D object, such as a 3D shape of the 3D input object, an albedo of the 3D input object, a pose of the 3D input object, and lighting illuminating the 3D input object. The physical properties are extracted from the 2D input image using a neural network that is trained to reconstruct the 2D input image. The 2D input image is reconstructed by disentangling the physical properties from pixels of the 2D input image using multiple subnetworks. The disentangled physical properties produced by the multiple subnetworks are combined into a 2D output image using a differentiable renderer.

Abstract: To control a hybrid dynamical system, a predictive feedback controller formulates a mixed-integer nonlinear programming (MINLP) problem including nonlinear functions of continuous optimization variables representing the continuous elements of the operation of the hybrid dynamical system and/or one or multiple linear functions of integer optimization variables representing the discrete elements of the operation of the hybrid dynamical system. The MINLP problem is formulated into a separable format ensuring that the discrete elements of the operation are present only in the linear functions of the MINLP problem. The MINLP problem is solved over multiple iterations using a partial convexification of a portion of a space of the solution including a current solution guess. The partial convexification produces a convex approximation of the nonlinear functions of the MINLP without approximating the linear functions of the MINLP to produce a partially convexified MINLP.

Abstract: A grating coupler having first and second ends for coupling a light beam to a waveguide of a chip includes a substrate configured to receive the light beam from the first end and transmit the light beam through the second end, the substrate having a first refractive index n1, a grating structure having curved grating lines arranged on the substrate, the grating structure having a second refractive index n1, wherein the curved grating lines have line width w and height d and are arranged by a pitch ?, wherein the second refractive index n2 is less than first refractive index n1, and a cladding layer configured to cover the grating structure, wherein the cladding layer has a third refractive index n3. The curves of the grating lines are constructed such that the emitting beam is shaped for efficient coupling to another optical component. The curves can also be tilted to reduce coupling back into the waveguide.

Abstract: A system for flexible regularization and adaptable scaling of an artificial neural network is provided.

Abstract: Systems and methods for a computer system for detecting anomalies in incoming communication from a sender to a receiver. Accepting a relationship structure defining a trained association model between the sender and the receiver, and the incoming communication. Accessing neural networks trained to detect anomalies in the incoming communication and classify the anomalies by type, subject to correspondence between content of the incoming communication and the trained association model between the sender and the receiver. Compute an updated association model, based on sender and the receivers organizational indications using the content of the incoming communication. Execute the neural networks by submitting the incoming communication and the updated association model to produce a result of anomaly detection and anomaly classification type.

Abstract: A computer-implemented learning method for optimizing a control policy controlling a system is provided. The method includes receiving states of the system being operated for a specific task, initializing the control policy as a function approximator including neural networks, collecting state transition and reward data using a current control policy, estimating an advantage function and a state visitation frequency based on the current control policy, updating the current control policy using the second-order approximation of the objective function, a second-order approximation of the KL-divergence constraint on the permissible change in the policy using a quasi-newton trust region policy optimization, and determining an optimal control policy, for controlling the system, based on the average reward accumulated using the updated current control policy.

Abstract: A control system for controlling a motion of a vehicle to a target driving goal uses a decision-maker configured to determine a sequence of intermediate goals leading to the next target goal by optimizing the motion of the vehicle subject to a first model and tightened driving constraints formed by tightening driving constraints by a safety margin, and uses a motion planner configured to determine a motion trajectory of the vehicle tracking the sequence of intermediate goals by optimizing the motion of the vehicle subject to the second model. The driving constraints include mixed logical inequalities of temporal logic formulae specified by traffic rules to define an area where the temporal logic formulae are satisfied, while the tightened driving constraints shrink the area by the safety margin, which is a function of a difference between the second model and the first model approximating the second model.

Abstract: A probabilistic system for tracking a state of a vehicle using unsynchronized cooperation of information includes a probabilistic multi-head measurement model relating incoming measurements with the state of the vehicle. The first head of the model relates measurements of the satellite signals subject to measurement noise with a belief on the state of the vehicle, and a second head relates an estimation of the state of the vehicle subject to estimation noise with the belief on the state of the vehicle. A probabilistic filter of the system updates recursively the belief on the state of the vehicle based on the multi-head measurement model accepting one or a combination of the measurements of the satellite signals subject to the measurement noise and the estimation of the state of the vehicle subject to the estimation noise.

Abstract: To control a hybrid dynamical system, a predictive feedback controller formulates a mixed-integer nonlinear programming (MINLP) problem including nonlinear functions of continuous optimization variables representing the continuous elements of the operation of the hybrid dynamical system and/or one or multiple linear functions of integer optimization variables representing the discrete elements of the operation of the hybrid dynamical system. The MINLP problem is formulated into a separable format ensuring that the discrete elements of the operation are present only in the linear functions of the MINLP problem. The MINLP problem is solved over multiple iterations using a partial convexification of a portion of a space of the solution including a current solution guess. The partial convexification produces a convex approximation of the nonlinear functions of the MINLP without approximating the linear functions of the MINLP to produce a partially convexified MINLP.

Abstract: A tracking system for tracking an expanded state of an object is provided. The tracking system executes, for a predetermined time period, a probabilistic filter that iteratively tracks a belief on the expanded state of the object, wherein the belief is predicted using a motion model of the object and is further updated using a compound measurement model of the object. After the predetermined time period, the updated beliefs are smoothed to generate a state-decoupled online batch of training data. The compound measurement model includes multiple probabilistic distributions constrained to lie around a contour of the object with a predetermined relative geometrical mapping to the center of the object. The compound measurement model is updated using the online batch of training data. Further, the tracking system tracks the expanded state of the object based on the updated compound measurement model.

Abstract: The present disclosure provides a system and a method for generating a radar image of a scene. The method comprises receiving radar measurements of a scene collected from a set of antennas, wherein the set of antennas are under uncertainties caused by one or a combination of position ambiguities and clock ambiguities of each of the antennas. The method further comprises generating the radar image of the scene by solving a sparse recovery problem. The sparse recovery problem determines, until a termination condition is met, a set of image shifts of the radar image corresponding to different uncertainties of the antennas and updates an estimate of the radar image, based on the determined set of image shifts of the radar image. The sparse recovery problem is solved with a neural network denoiser that denoises a filtering of the estimate of the radar image.

Abstract: A train trip controller system is provided. The train trip controller system may collect operational data for the train, the operational data comprising at least: an itinerary information for a trip the train, constraints on capacity of the train and on ratio between number of passenger cars and number of seats in a particular type of passenger car for different types of passenger cars, an operational cost factor for adding and removing passenger cars, sales horizon condition, and a congestion factor. Based on the operational data, the train trip controller system is configured to determine an upper bound of a stochastic cost function, using a mean of arrival rate of passengers over the sales horizon condition, for each of a different type passenger car. The computation of the upper bound is then used to determine a ticket price and capacity to be used for the train. The determined ticket price and capacity are in turn used to achieve an optimization objective for the train trip controller system.

Abstract: A permittivity sensor, for determining an image of a distribution of permittivity of a material of an object in a scene, comprising an input interface, a hardware processor, and an output interface is provided. The input interface is configured to accept phaseless measurements of propagation of a known incident field through the scene and scattered by the material of the object in the scene. The hardware processor is configured to solve a multi-variable minimization problem over unknown phases of the phaseless measurements and unknown image of the permittivity of the material of the object by minimizing a difference of a nonlinear function of the known incident field and the unknown image with a product of known magnitudes of the phaseless measurements and the unknown phases. Further, the output interface is configured to render the permittivity of the material of the object provided by the solution of the multi-variable minimization problem.

Abstract: A controller and a method for controlling motion of a vehicle is provided. The method comprises acquiring motion information including a current state of the vehicle and a desired state of the vehicle, determining a combination of a steering angle of the wheels and motor forces for moving the vehicle from the current state into the desired state by using a first model of the motion of the vehicle and a second model of the motion of the chassis of the vehicle, determining a cost function of the motion of the vehicle, optimizing the cost function of the motion of the vehicle to compute a command signal for controlling the steering wheel and the plurality of electric motors, and controlling the steering angle of the wheels and the motor forces based on the command signal.

Abstract: Embodiments of the present disclosure disclose a method and a system for controlling a motion of an electric vehicle (EV). The method includes determining a velocity profile moving the EV from an initial velocity over a period of time by minimizing the energy dissipation according to an energy-loss function. The energy-loss function maps values of acceleration and velocity of the EV to energy dissipation of the EV resulting from controlling one or multiple electric motors of the EV to move the EV at corresponding acceleration and velocity values. The velocity profile is a function of time. The method further includes controlling the one or multiple electric motors of the EV to generate a torque for moving the EV according to the velocity profile.

Abstract: A method for controlling a system by a controller comprises accepting a current state of the system and selecting, using a trained function of the current state, a solver from a set of solvers. The method further comprises solving an optimal control optimization problem using the selected solver to produce a current control input, such that for at least some different control steps, the predictive controller solves a formulation of the optimal control optimization problem with different solvers having different accuracies, requiring different computational resources, or both and submitting the current control input to the system thereby changing the current state of the system.

Abstract: A system for detecting an anomaly in a video of a factory automation scene is disclosed. The system may accept the video; accept a set of training feature vectors derived from spatio-temporal regions of a training video, where a spatio-temporal region is associated with one or multiple training feature vectors; partition the video into multiple sequences of video volumes; produce a sequence of binary difference images for each of the video volumes; count occurrences of each of predetermined patterns of pixels in each binary difference image for each of the video volumes to produce an input feature vector including an input motion feature vector defining a temporal variation of counts of the predetermined patterns for each of the video volumes; produce a set of distances based on the produced input feature vectors and the set of training feature vectors; and detect the anomaly based on the produced set of distances.

Abstract: Embodiments of the present disclosure disclose a method and a system for training a neural network for generating universal adversarial perturbations. The method includes collecting a plurality of data samples. Each of the plurality of data samples is identified by a label from a finite set of labels. The method includes training a probabilistic neural network for transforming the plurality of data samples into a corresponding plurality of perturbed data samples having a bounded probability of deviation from the plurality of data samples by maximizing a conditional entropy of the finite set of labels of the plurality of data samples conditioned on the plurality of perturbed data samples. The conditional entropy is unknown. The probabilistic neural network is trained based on an iterative estimation of a gradient of the unknown conditional entropy of labels. The method further includes generating the universal adversarial perturbations based on the trained probabilistic neural network.

Abstract: Present disclosure discloses an image processing system and method for manipulating two-dimensional (2D) images of three-dimensional (3D) objects of a predetermined class (e.g., human faces). A 2D input image of a 3D object of the predetermined class is manipulated by manipulating physical properties of the 3D object, such as a 3D shape of the 3D input object, an albedo of the 3D input object, a pose of the 3D input object, and lighting illuminating the 3D input object. The physical properties are extracted from the 2D input image using a neural network that is trained to reconstruct the 2D input image. The 2D input image is reconstructed by disentangling the physical properties from pixels of the 2D input image using multiple subnetworks. The disentangled physical properties produced by the multiple subnetworks are combined into a 2D output image using a differentiable renderer.

Abstract: A wireless smart utility network (Wi-SUN) device participating in a Wi-SUN network for coexistence with a Wi-Fi HaLow network sharing frequency spectra between the networks is provided.