Abstract: System and method for robust machine learning (ML) includes an attack detector comprising one or more deep neural networks trained using adversarial examples generated from a generative adversarial network (GAN), producing an alertness score based on a likelihood of an input being adversarial. A dynamic ensemble of individually robust ML models of various types and sizes and all being trained to perform an ML-based prediction is dynamically adapted by types and sizes of ML models to be deployed during the inference stage of operation. The adaptive ensemble is responsive to the alertness score received from the attack detector. A data protector module with interpretable neural network models is configured to prescreen training data for the ensemble to detect potential data poisoning or backdoor triggers in initial training data.

Abstract: System and method are disclosed for approximating unknown safety constraints during reinforcement learning of an autonomous agent. A controller for directing the autonomous agent includes a reinforcement learning (RL) algorithm configured to define a policy for behavior of the autonomous agent, and a control barrier function (CBF) algorithm configured to calculate a corrected policy that relocates policy states to an edge of a safety region. Iterations of the RL algorithm safely learn an optimal policy where exploration remains within the safety region. CBF algorithm uses standard least squares to derive estimates of coefficients for linear constraints of the safe region. This overcomes inaccurate estimation of safety region constraints caused by one or more noisy observations of constraints received by sensors.

Abstract: A computer-implemented method of scheduling jobs for an industrial process includes receiving jobs to be executed on machines within a manufacturing facility. A job schedule is generated based on an optimization function that minimizes total energy cost for all the machines during a time horizon based on a summation of energy cost at each time step between a start time and an end time. The energy cost at each time step is a summation of (a) a first energy cost associated with each machine in sleeping mode during the time step, (b) a second energy cost associated with each machine in stand-by mode during the time step, and (c) a third energy cost associated with each machine in processing mode during the time step. The jobs are executed on the machines based on the job schedule.

Abstract: A computer-implemented method for training a deep neural network includes defining a loss function corresponding to the deep neural network, receiving a training dataset comprising training samples, and setting current parameter values to initial parameter values. An optimization method is performed which iteratively minimizes the loss function. During each iteration, a steepest direction of the loss function is calculated by determining the gradient of the loss function at the current parameter values. A batch of samples included in training samples is selected. A matrix-free CG solver is applied to obtain an inexact solution to a linear system defined by the steepest direction of the loss function and a stochastic Hessian matrix with respect to the batch of samples. A descent direction is determined, and the parameter values are updated based on the descent direction. Following the optimization method, the parameter values are stored in relationship to the deep neural network.

Abstract: Systems, techniques, and computer-program products are provided to generate manufacturing schedules that integrate maintenance strategies. A manufacturing schedule can be generated by solving an optimization problem subject to operational constraints that preserve consistency in the order of the operations to be performed during the manufacture of a product, and further subject to maintenance constraints that enforce a desired maintenance strategy. The optimization problem can be solved by minimizing a makespan of a product subject to the operational and maintenance constraints.

Abstract: A computer-implemented method of scheduling jobs for an industrial process includes receiving jobs to be executed on machines within a manufacturing facility. A job schedule is generated based on an optimization function that minimizes total energy cost for all the machines during a time horizon based on a summation of energy cost at each time step between a start time and an end time. The energy cost at each time step is a summation of (a) a first energy cost associated with each machine in sleeping mode during the time step, (b) a second energy cost associated with each machine in stand-by mode during the time step, and (c) a third energy cost associated with each machine in processing mode during the time step. The jobs are executed on the machines based on the job schedule.

Abstract: A method for training a deep learning network includes defining a loss function corresponding to the network. Training samples are received and current parameter values are set to initial parameter values. Then, a computing platform is used to perform an optimization method which iteratively minimizes the loss function. Each iteration comprises the following steps. An eigCG solver is applied to determine a descent direction by minimizing a local approximated quadratic model of the loss function with respect to current parameter values and the training dataset. An approximate leftmost eigenvector and eigenvalue is determined while solving the Newton system. The approximate leftmost eigenvector is used as negative curvature direction to prevent the optimization method from converging to saddle points. Curvilinear and adaptive line-searches are used to guide the optimization method to a local minimum. At the end of the iteration, the current parameter values are updated based on the descent direction.

Abstract: The Support Vector Machine (SVM) has been used in a wide variety of classification problems. The original SVM uses the hinge loss function, which is nondifferentiable and makes the problem difficult to solve in particular for regularized SVMs, such as with l1-norm. The Huberized SVM (HSVM) is considered, which uses a differentiable approximation of the hinge loss function. The Proximal Gradient (PG) method is used to solving binary-class HSVM (BHSVM) and then generalized to multi-class HSVM (MHSVM). Under strong convexity assumptions, the algorithm converges linearly. A finite convergence result about the support of the solution is given, based on which the algorithm is further accelerated by a two-stage method.

Abstract: A computer-implemented method for performing predictive maintenance includes executing a fleet prediction process. During this fleet prediction process, a plurality of fleet data records is collected. Each fleet data record comprises sensor data from a particular physical component in a fleet of physical components. A plurality of component maintenance predictions related to the fleet of physical components is generated. Each component maintenance prediction corresponds to a particular physical component. The plurality of component predictions are merged into one or more fleet maintenance predictions and the fleet maintenance predictions are presented to one or more users. Following the fleet prediction process, a next execution of the fleet prediction process is scheduled based on the fleet maintenance predictions.

Abstract: The present embodiments relate to reducing the input dimensions to a machine-based Bayesian Optimization using stacked autoencoders. By way of introduction, the present embodiments described below include apparatuses and methods for pre-processing a digital input to a machine-based Bayesian Optimization to a lower the dimensional space of the input, thereby lowering the bounds of the Bayesian optimization. The output of the Bayesian Optimization is then projected back into the original dimensional space to determine input and output values in the original dimensional apace. As such, the optimization is performed by the machine in a lower dimension using the stacked autoencoder to constrain the input dimensions to the optimization.

Abstract: A method for training a deep learning network includes defining a loss function corresponding to the network. Training samples are received and current parameter values are set to initial parameter values. Then, a computing platform is used to perform an optimization method which iteratively minimizes the loss function. Each iteration comprises the following steps. An eigCG solver is applied to determine a descent direction by minimizing a local approximated quadratic model of the loss function with respect to current parameter values and the training dataset. An approximate leftmost eigenvector and eigenvalue is determined while solving the Newton system. The approximate leftmost eigenvector is used as negative curvature direction to prevent the optimization method from converging to saddle points. Curvilinear and adaptive line-searches are used to guide the optimization method to a local minimum. At the end of the iteration, the current parameter values are updated based on the descent direction.

Abstract: A method (100) for electricity demand shaping through load shedding and shifting in an electrical smart grid.

Abstract: A computer-implemented method for determining optimized amount of radiopharmaceutical to be produced at a production facility, the radiopharmaceutical being for use in nuclear imaging at customer sites, in order to meet aggregate demands of orders placed by the customer sites (e.g. medical imaging centers, hospitals, etc.), wherein the quantity of radiopharmaceutical is sufficient to meet the aggregate demand while minimizing any overproduction of the radiopharmaceutical.

Abstract: A method (100) for demand shaping through load shedding and shifting in an electrical smart grid.

Abstract: A method to manage operating costs of a combined cooling heating and power (CCHP) plant that includes converting complex models of underlying components of the plant into simplified models (S101), performing an optimization that uses the simplified models as constraints of the optimization to output at least one decision variable (S102), and adjusting controls of the plant based on one or more of the output decision variables (S103).

Abstract: A system and method of predicting future power plant operations is based upon an artificial neural network model including one or more hidden layers. The artificial neural network is developed (and trained) to build a model that is able to predict future time series values of a specific power plant operation parameter based on prior values. By accurately predicting the future values of the time series, power plant personnel are able to schedule future events in a cost-efficient, timely manner. The scheduled events may include providing an inventory of replacement parts, determining a proper number of turbines required to meet a predicted demand, determining the best time to perform maintenance on a turbine, etc. The inclusion of one or more hidden layers in the neural network model creates a prediction that is able to follow trends in the time series data, without overfitting.

Abstract: A generalized autoregressive integrated moving average (ARIMA) model for use in predictive analytics of time series is based upon creating all possible ARIMA models (by knowing a priori the largest possible values of the p, d and q parameters forming the model), and utilizing the results of at least two different performance measures to ultimately choose the ARIMA(p,d,q) model that is most appropriate for the time series under study. The method of the present invention allows each parameter to range over all possible values, and then evaluates the complete universe of all possible ARIMA models based on these combinations of p, d and q to find the specific p, d and q parameters that yield the “best” (i.e., lowest value) performance measure results. This generalized ARIMA model is particularly useful in predicting future operating hours of power plants and scheduling maintenance events on the gas turbines at these plants.

Abstract: A method (100) for demand shaping through load shedding and shifting in an electrical smart grid.

Abstract: A method (100) for electricity demand shaping through load shedding and shifting in an electrical smart grid.

Abstract: The Support Vector Machine (SVM) has been used in a wide variety of classification problems. The original SVM uses the hinge loss function, which is nondifferentiable and makes the problem difficult to solve in particular for regularized SVMs, such as with l1-norm. The Huberized SVM (HSVM) is considered, which uses a differentiable approximation of the hinge loss function. The Proximal Gradient (PG) method is used to solving binary-class HSVM (BHSVM) and then generalized to multi-class HSVM (MHSVM). Under strong convexity assumptions, the algorithm converges linearly. A finite convergence result about the support of the solution is given, based on which the algorithm is further accelerated by a two-stage method.