Abstract: A diagnostic device and diagnostic method for monitoring the operation of a control loop with a controlled system having a valve as an actuator, wherein in the case of a substantially stepped profile of a setpoint, a linear submodel is identified, which is subsequently used, with a substantially constant setpoint, in order to identify a nonlinear submodel, which is arranged upstream of the linear submodel in a Hammerstein model for the controlled system so as to facilitate a quantitative assessment of the friction behavior of a valve with a comparatively low level of computing complexity.

Abstract: Systems, methods, and apparatus, including computer programs encoded on a computer storage medium for processing a network input through a neural network having one or more initial neural network layers followed by a softmax output layer. In one aspect, the methods include obtaining a layer output generated by the one or more initial neural network layers and processing the layer output through the softmax output layer to generate a neural network output. Processing the layer output through the softmax output layer includes determining, for each possible output value, a number of occurrences in the layer output values; for each possible output value occurring in the layer output values, determining a respective exponentiation measure; determining a normalization factor for the layer output by combining the exponentiation measures in accordance with the number of occurrences of the possible output values; and determining, for each of layer output values, a softmax probability value.

Abstract: Disclosed are methods, apparatus and systems for improving data management and workload distribution in pattern recognition systems. An example method of managing data for a sign language translation system includes receiving multiple sets of data acquired by one or more data acquisition devices. Each set of data including an image frame that illustrates at least a part of a gesture. The method includes determining, for each of the multiple sets of data, a plurality of attribute values defined by a customized template. The method includes accessing the multiple sets of data, by a plurality of processing units, based on a location indicated by the attributes for recognizing the at least a part of a gesture.

Abstract: Some embodiments described herein cover a machine learning architecture with a separated perception subsystem and application subsystem. These subsystems can be co-trained to yield a lifelong learning system that can capture spatio-temporal regularities in its inputs. In one example embodiment a first node of the machine learning architecture receives a data item. The first node receives a first feature vector that was generated at a first time. The first node processes information from at least a portion of the data item and at least a portion of the first feature vector at a second time to generate a second feature vector. The first node generates a sparse feature vector from the second feature vector, wherein a majority of feature elements in the sparse feature vector have a value of zero. A second node of the machine learning architecture then processes the sparse feature vector to determine a first output.

Abstract: Methods and systems represent constraint as an Ising model penalty function and a penalty gap associated therewith, the penalty gap separating a set of feasible solutions to the constraint from a set of infeasible solutions to the constraint; and determines the Ising model penalty function subject to the bounds on the programmable parameters imposed by the hardware limitations of the second processor, where the penalty gap exceeds a predetermined threshold greater than zero. Such may be employed to find quantum binary optimization problems and associated gap values employing a variety of techniques.

Abstract: Identify individual machines of a multi-machine computing system. Construct a graph of dependencies among the machines. Obtain estimated total administration times and administration priorities for each of the machines. Identify availability of administration resources to assist in administration of one or more of the machines. Select a first set of machines for administration in response to the graph, administration priorities, estimated total administration times, and availability of the first set of administration resources, and administer the first set of machines in parallel using the first set of administration resources. Update the graph in response to administration of the first set of machines. Select a subsequent set of machines for administration in response to the updated graph, administration priorities, estimated total administration times, and availability of a subsequent set of administration resources.

Abstract: The present invention provides a method and system for querying a trie data structure. The method and system include constructing a pool of linear programming problems for a trie data structure, storing the pool of linear programming problems in a memory, receiving a request, setting variables in one linear programming problem to unity, sending the one linear programming problem to a solver, wherein the solver sets a p variable corresponding to each path to 0 or 1, iterating through p variables corresponding to one or more paths, considering at most n paths, selecting one path from the n paths, assigning same values as that of one or more variables of the unknown value type nodes that are present in the selected one path, and reselling variables in the one linear programming problem to zero that was previously set to unity.

Abstract: An apparatus for detecting an object in an image, including a detection group determiner configured to classify a plurality of feature levels into one or more detection groups, wherein the plurality of feature levels includes multiple scales and are arranged as a feature pyramid; and a detection group processor configured to create one or more feature maps corresponding to the one or more detection groups, and to detect an object by applying one or more models to the one or more feature maps.

Abstract: Technologies are generally described for an array of logic elements effective to generate a data signal. A first logic element may include an input circuit, a comparator circuit, and a state machine. The input circuit may receive a first input state signal and a second input state signal from a second logic element and a third logic element, respectively. The input circuit may determine a sum based on the first and second input state signals. The comparator circuit may compare the sum with a threshold and, in response, may generate an intermediate signal based on the comparison. The state machine may identify a current state of the first logic element. The state machine may generate an output state signal based on the intermediate signal and the current state of the first logic element. The output state signal may indicate a subsequent state of the first logic element.

Abstract: A method to associate a set of first entities to a set of second entities, e.g., computing jobs to processors, agent teams to workspace resources within a physical location, or the like. The NG is seeded using a force directed graph (FDG), whose “seed” particles represents the agents and their relative interconnectedness. The FDG is first brought into an equilibrium state to define a solution space. A relative coordinate system of the FDG solution space is then translated to a number of vertices represented in the NG, and then an initial seeding of the seed particles in the NG (based on their relative positions in the FDG solution space) is carried out. A search is then performed. During the search, each seed vertex releases its embedded agents to adjacent vertices to enable the agents to search for and achieve a required count.

Abstract: The present embodiments relate to detecting multiple landmarks in medical images. By way of introduction, the present embodiments described below include apparatuses and methods for detecting landmarks using hierarchical feature learning with end-to-end training. Multiple neural networks are provided with convolutional layers for extracting features from medical images and with a convolutional layer for learning spatial relationships between the extracted features. Each neural network is trained to detect different landmarks using a different resolution of the medical images, and the convolutional layers of each neural network are trained together with end-to-end training to learn appearance and spatial configuration simultaneously. The trained neural networks detect multiple landmarks in a test image iteratively by detecting landmarks at different resolutions, using landmarks detected a lesser resolutions to detect additional landmarks at higher resolutions.

Abstract: A computer program product comprising a non-transitory computer readable storage medium retaining program instructions configured to cause a processor to perform actions, which program instructions implement: a framework for creating a model of an NP-hard problem, the model comprising at least one entity selected from the group comprising: an objective, a variable, an equation and a constraint, wherein the framework provides methods for automatically transforming the model, comprising: one or more methods for manipulating or changing a status of the entity of the model, the methods comprising a method for imposing or ignoring the constraint; and one or more methods related to operations to be applied to the entity of the model.

Abstract: A system, method and computer program product are provided for combining raw data from multiple metrology tools. Reference values are obtained for at least one parameter of a training component. Signals are collected for the at least one parameter of the training component, utilizing a first metrology tool and a different second metrology tool. Further, at least a portion the signals are transformed into a set of signals, and for each of the at least one parameter of the training component, a corresponding relationship between the set of signals and the reference values is determined and a corresponding training model is created therefrom. Signals from a target component are collected utilizing at least the first metrology tool and the second metrology tool, and each created training model is applied to the signals collected from the target component to measure parametric values for the target component.

Abstract: A neural network system for identifying positions of objects in an input image can include an object detector neural network, a memory interface subsystem, and an external memory. The object detector neural network is configured to, at each time step of multiple successive time steps, (i) receive a first neural network input that represents the input image and a second neural network input that identifies a first set of positions of the input image that have each been classified as showing a respective object of the set of objects, and (ii) process the first and second inputs to generate a set of output scores that each represents a respective likelihood that an object that is not one of the objects shown at any of the positions in the first set of positions is shown at a respective position of the input image that corresponds to the output score.

Abstract: A method for performing optical constellation conversion, according to which each received symbol from a constellation of input symbols is optically split into M components and each component is multiplied by a first predetermined different complex weighing factor, to achieve M firstly weighted components with different amplitudes. Then a nonlinear processor optically performs a nonlinear transform on each M firstly weighted components, so as to obtain M outputs which are linearly independent, Finally, a linear processor optically performs a linear transform to obtain a new converted constellation by optically multiplying, in the complex plane, each of the M outputs by a second predetermined different complex weighing factor, to achieve M secondly weighted components and then summing the M secondly weighted components.

Abstract: The creation and updating of an association index that defines a linking between a child table and a parent table in a database system, and in which each of at least some of the parent table rows have an associated expression defining rules for mapping child table rows to the associated parent table row. The association index may be constructed with one pass of the parent table by evaluating the mapping definition to identifying associated child table rows. If there are such associations for a given parent table row, the linking module may record that association in an association index. If there are changes made that potentially invalidate an association status (whether an association or lack thereof), a re-evaluation module then determines which mapping definitions for which parent rows need to be re-evaluated, and potentially for which child table rows, rather than perform the entire process again.

Abstract: A method and system are disclosed for solving a convex integer quadratic programming problem using a binary optimizer, the method comprising use of a processor for receiving a convex integer quadratic programming problem; converting the convex integer quadratic programming problem into a plurality of constrained and unconstrained binary quadratic programming problems and providing the plurality of unconstrained binary quadratic programming problems to the binary optimizer to thereby solve the convex integer quadratic programming problem.

Abstract: A method for generating a system for predicting saliency in an image and method of use of the prediction system are described. Attention maps for each of a set of training images are used to train the system. The training includes passing the training images though a neural network and optimizing an objective function over the training set which is based on a distance measure computed between a first probability distribution computed for a saliency map output by the neural network and a second probability distribution computed for the attention map for the respective training image. The trained neural network is suited to generation of saliency maps for new images.

Abstract: In one exemplary embodiment, a computer-implemented method includes obtaining a data set from a data source. The data set is prepared for an analysis operation according to a problem type. A result is generated from an interactive geometric node based a geometric property of the data set. A specified condition is determined with the result from the interactive geometric node based on a query to the interactive geometric node. A system's geometric configuration is determined. A geometric control criterion is determined.

Abstract: Various embodiments select features from a feature space. In one embodiment, a set of training samples and a set of test samples are received. A first centered Gram matrix of a given dimension is determined for each of a set of feature vectors that include at least one of the set of training samples and at least one of the set of test samples. A second centered Gram matrix of the given dimension is determined for a target value vector that includes target values from the set of training samples. A set of columns and rows associated with the at least one of the test samples in the second centered Gram matrix is set to 0. A subset of features is selected from a set of features based on the first and second centered Gram matrices.

Abstract: A method to optimize workload across a plurality of storage devices of a storage system, where the method monitors a workload of a first storage device belonging to a first tier of the storage system, calculates a performance of the workload of the first storage device belonging to a first tier of the storage system, interpolates a performance threshold for the first storage device using the workload pattern of the first storage device and a profile of the first storage device, the profile identifying a benchmark performance of the first storage device, and optimizes a usage of the first storage device within the storage system to improve a performance of the first storage device.

Abstract: Dynamic parameter tuning using particle swarm optimization is disclosed. According to one embodiment, a system for dynamically tuning parameters comprising a control unit; and a system for receiving parameters tuned by the control unit. The control unit receives as input a model selection and definitions, and dynamically tunes a value for each parameter by using a modified particle swarm optimization method. The modified particle swarm optimization method comprises moving particle locations based on a particle's inertia, experience, global knowledge, and a tuning factor. The control unit outputs the dynamically tuned value for each parameter.

Abstract: An intelligent control system based on an explicit model of cognitive development (Table 1) performs high-level functions. It comprises up to O hierarchically stacked neural networks, Nm, . . . , Nm+(O?1), where m denotes the stage/order tasks performed in the first neural network, Nm, and O denotes the highest stage/order tasks performed in the highest-level neural network. The type of processing actions performed in a network, Nm, corresponds to the complexity for stage/order m. Thus N1 performs tasks at the level corresponding to stage/order 1. N5 processes information at the level corresponding to stage/order 5. Stacked neural networks begin and end at any stage/order, but information must be processed by each stage in ascending order sequence. Stages/orders cannot be skipped. Each neural network in a stack may use different architectures, interconnections, algorithms, and training methods, depending on the stage/order of the neural network and the type of intelligent control system implemented.

Abstract: An approach is provided to receive audible speech and convert the received speech to text while the audible speech is being delivered to a user. An annotation candidate is identified in the text and an annotation reference relating to the identified annotation candidate is retrieved and presented to the user.

Abstract: In one embodiment, network data is received at a first node in a computer network. A low precision machine learning model is used on the network data to detect a network event. A notification is then sent to a second node in the computer network that the network event was detected, to cause the second node to use a high precision machine learning model to validate the detected network event.

Abstract: A method system and computer program product are disclosed for identifying silent failures in an application, comprising of generating a finite state machine (FSM) model of the application based on an input data, extracting state specific invariants relevant to the generated FSM, performing a localized invariant violation check at each state of the generated FSM and upon detection of an invariant violation at any state of the FSM, logging the violation as a silent failure.

Abstract: Techniques for profiling input/output (I/O) characteristics in a computing system implemented with distributed infrastructure. In one example, a method comprises the following steps. Behavioral characteristics are extracted for a set of factors associated with a targeted system in a distributed infrastructure by causing execution of one or more input/output workloads in accordance with one or more entities and the targeted system. At least a portion of the extracted behavioral characteristics is utilized to determine whether one or more subsequent workloads satisfy one or more quality-of-service criteria when executed in accordance with at least one of the one or more entities and the targeted system.

Abstract: In one exemplary embodiment, a computer-implemented method includes obtaining a data set from a data source. The data set is prepared for an analysis operation according to a problem type. A result is generated from an interactive geometric node based a geometric property of the data set. A specified condition is determined with the result from the interactive geometric node based on a query to the interactive geometric node. A system's geometric configuration is determined. A geometric control criterion is determined.

Abstract: Embodiments of the present invention provide a workload optimization approach that measures workload performance across combinations of hardware (platform, network configuration, storage configuration, etc.) and operating systems, and which provides a workload placement on the platforms where jobs perform most efficiently. This type of placement may be based on performance measurements (e.g., throughput, response, and other such service levels), but it can also be based on other factors such as power consumption or reliability. In a typical embodiment, ideal platforms are identified for handling workloads based on performance measurements and any applicable service level agreement (SLA) terms.

Abstract: An ad-hoc service provider is configured to support a wireless connection to the network. The ad-hoc service provider is further configured to provide access to the network for one or more mobile clients. The one or more mobile clients are configured to select the ad-hoc service provider based on at least one parameter related to the ad-hoc service provider's ability to support the wireless connection. The ad-hoc service provider is further configured to allocate bandwidth to the one or more mobile clients based on at least one parameter related to the effect on the ad-hoc service provider for providing such bandwidth.

Abstract: Systems, methods, and devices of the various embodiments enable graph based routing of containers and adjustments of container routing based on one or more current material handling system state. The various embodiments enable examination and evaluation of the material handling systems using graphs to find an optimal route from a starting node in the material handling system to an ending node in the material handling system. In an embodiment, the graph may be a directed graph of edges and nodes of a material handling system. In an embodiment, the directed graph may include edge weights and the edge weights may be used to determine the optimal route for the container. In an embodiment, the edge weights may be updated as events occur in the material handling system which may result in dynamic routing of containers through the material handling system.

Abstract: Described is a system for multispectral image processing with spiking dynamics. For example, the system receives an input image and compresses the image through space and spectrally variant sampling. Center-surround dynamics are modeled to control high dynamic ranges of the image and provide gain control. Further, habituative dynamics are modeled to produce outputs specialized for static or dynamic image content. Finally, neural spikes are generated based on the habituative dynamics. The neural spikes are saved or provided to other systems for further image processing.

Abstract: An optimal path search system, which optimizes a travel path of a moving object, comprising: a path information database; and a path search unit that searches for an optimal path that exhibits a minimum travel cost borne by the moving object, from among travel paths of the moving object, wherein: the path search unit is configured to: determine destination points on the path network; search for at least one nearest node to each of destination points, search for a second travel path along which the moving object travels by passing through the at least one nearest node and all of the at least one destination point, determine the second travel path for which a travel cost is minimum; and search for the travel paths followed by using each of a plurality of moving means, respectively, in a case where the moving object travels along the optimal path.

Abstract: According to an embodiment of the present invention, a system optimizes an information processing environment, and comprises at least one processor. The system collects information pertaining to operational behavior of the information processing environment and including a plurality of parameters. A neural network structure is established to associate the parameters to a desired operational performance characteristic for the information processing environment. The neural network structure is trained with the collected information from the information processing environment to produce a model for the information processing environment. The model is optimized to determine values for the parameters and the information processing environment is adjusted based on the determined parameter values to attain the desired operational performance of the information processing environment.

Abstract: Provided is an abnormality detecting device for a construction machine that can estimate an abnormality occurring to a component (engine, pump, etc.) of the construction machine based on the relationship among a plurality of pieces of sensor information and thereby prevent machine failure. A correlation coefficient calculation unit 102 calculates correlation coefficients between time-series sensor values acquired by a plurality of sensors 101. A correlation coefficient comparison unit 103 compares the correlation coefficients and calculates the degree of difference between each correlation coefficient and other correlation coefficients. An abnormality judgment unit 104 judges that an abnormality has occurred to a part related to a sensor when the degree of difference calculated in regard to the sensor exceeds a preset value.

Abstract: In accordance with certain embodiments of the present disclosure, a regionalization method is disclosed. The method includes inputting a data set into a computer. The method further includes utilizing the computer to perform contiguity-constrained hierarchical clustering on the data set to generate two regions and performing a fine-tuning procedure on the two regions with the computer to iteratively modify the boundaries between the two regions.

Abstract: A method and apparatus are presented for automatically fabricating arbitrary materials and objects from raw components, using a combination of simple chemical, electrical, and mechanical operations. The invention will automatically generate machine control instructions for controlling automated fabrication devices and equipment from simple instructions in natural language. The invention also allows the sharing, remote execution, scheduling, and automatic ingredient ordering for such instructions to allow the creation of new materials and professional object fabrication with little or no human intervention.

Abstract: A method of model identification for a process with unknown initial conditions in an industrial plant, the method comprising collecting a set of manipulated variables and corresponding set of process variables from the process; obtaining a plurality of manipulated variables from the collected set of manipulated variables; for each of the plurality of manipulated variables, obtaining optimal model parameters of a model transfer function and computing a model fitting index for optimized simulated process variables generated by the model transfer function using the optimal model parameters; identifying a best model fitting index among the model fitting indices computed; selecting a manipulated variable associated with the best model fitting index as an initial steady state condition for the model transfer function; and selecting the optimal model parameters corresponding with the best model fitting index as the best model parameters of the model transfer function to tune the controller.

Abstract: Robust content-based decision-making support is enabled by software with a customizable knowledge base. Utilizing proprietary information contained within a knowledge base, the software enables users to search the indexed database by feature, example firm, or pattern and update the knowledge base based on the results. The information contained in the knowledge base enables results to be ranked by relevance and enables other feedback to be provided. The system and methods provide process support by helping financial professionals identify, analyze, and construct data analysis patterns based on individual domain knowledge and preferences. The system and methods automatically detect abnormal patterns and automatically analyze their correlations to market events to provide further process support to financial professionals. Using the results of any searching, analysis, and processing, the system and methods provide a neural network or other learning algorithm to provide content-based decision-making support.

Abstract: A method for providing trajectory planning for an aircraft based on constraint processing is disclosed. The method may be performed utilizing a computer or processor onboard the aircraft that is en route, through all phases of flight, to an end location. The trajectory planning method may comprise receiving a set of constraints during the flight of the aircraft; analyzing the set of constraints to determine an optimal trajectory between a current location of the aircraft and the end location, the optimal trajectory is determined based on compliance with the set of constraints; and dynamically adjusting the flight of the aircraft based on the optimal trajectory.

Abstract: A method for acquiring an input value such that an output value differs before and after refactoring is not known and an information processing apparatus searching for such an input value that inequivalent output values can be brought about for the equivalent input values among multiple target programs. The apparatus including: an acquisition section acquiring, for each of the multiple target programs, an input/output constraint showing a condition to be satisfied by an input value and an output value; a constraint condition generating section generating a constraint condition which becomes true when the multiple input/output constraints for the multiple target programs are satisfied, input values for the multiple target programs are equivalent, and output values for the multiple target programs are not equivalent; and a constraint releasing section giving the constraint condition to a constraint solver to obtain an input value satisfying the constraint condition.

Abstract: In various embodiments, electronic apparatus, systems, and methods include a unified compact spatiotemporal method that provides a process for machines to deal with space and time and to deal with sensors and effectors. Additional apparatus, systems, and methods are disclosed.

Abstract: A quantified belief propagation (QBP) algorithm receives as input an existentially quantified boolean formula (QBF) of existentially quantified boolean variables, universally quantified variables, and boolean operators. A tripartite graph is constructed, and includes (i) there-exists nodes that correspond to and represent the existentially quantified variables, (ii) for-all nodes that correspond to and represent the universally quantified variables, and (iii) sub-formula nodes that correspond to and represent sub-formulas of the QBF. A set of boolean values of the existentially quantified variables is found by (i) passing a first message from an arbitrary sub-formula node to an arbitrary for-all node, and (ii) in response, passing a second message from the arbitrary for-all node to the arbitrary sub-formula node.

Abstract: A method for solving the reconfiguration planning problem transforms the requirement that an invariant is always preserved into a constraint on the times at which the configuration parameters change. The method then solves this constraint to obtain the required reconfiguration plan. When this constraint is strengthened with a synthesis constraint, it is ensured that any final configuration that is computed is also reachable. An extension of the method allows parameters to take on multiple intermediate values.

Abstract: According to an embodiment of the present invention, a system optimizes an information processing environment, and comprises at least one processor. The system collects information pertaining to operational behavior of the information processing environment and including a plurality of parameters. A neural network structure is established to associate the parameters to a desired operational performance characteristic for the information processing environment. The neural network structure is trained with the collected information from the information processing environment to produce a model for the information processing environment. The model is optimized to determine values for the parameters and the information processing environment is adjusted based on the determined parameter values to attain the desired operational performance of the information processing environment.

Abstract: A solver solves constraint satisfaction problems. The solver receives a user decision and propagates the user decision over a constraint satisfaction problem network. The solver then determines if there is a conflict on the network based on the propagation. If there is a conflict, the solver allows a user to resolve the conflict interactively by either canceling the user decision or dropping one or more previously made user decisions.

Abstract: The approaches described herein provide an efficient way to process certain kinds of queries that retrieve XML data stored in an object-relational database. An XML query may include a predicate condition that joins two row sources. One or both of the row sources required for predicate evaluation may be path-based; that is, consume and produce rows of XMLType data. Such a path-based row source may be modified to use the predicate condition to filter the rows it generates so that only rows satisfying the predicate condition need to be further processed. A small cache structure introduced into the path-based row source may be used to optimize the evaluation of the predicate row. Once populated with row data, the cache structure may be reorganized to optimize the performance of predicate evaluation.

Abstract: A learning and training process that memorizes traces of a learner solving an exercise, attaches meta labels to the learner traces, and compares the learner traces to reference traces of a referent solving the exercise, meta labels being attached to the reference traces.

Abstract: A method for optimizing expensive functions with expensive nonlinear constraints. The method includes selecting sample data for evaluating an expensive function of a simulation, generating a function proxy model for the expensive function and a constraint proxy model for an expensive nonlinear constraint of the expensive function using an approximation scheme, calculating a first solution point for the simulation using the proxy models, and evaluating the expensive function at the first solution point using the sample data. When the expensive function and the proxy models do not converge at the first solution point, the method further includes adding the first solution point to the sample data for updating the proxy models. The method further includes repeating the calculation and evaluation of solution points until the expensive function and the proxy models converge and, following convergence, identifying an optimal solution of the function proxy model and the constraint proxy model.

Abstract: A computer-implemented land planning system is designed to generate at least one conceptual fit solution to a user-defined land development problem. The system electronically creates at least one candidate solution to the land development problem. The candidate solution incorporates a number of engineering measurements applicable in development of an undeveloped land site. A fitness function quantitatively evaluates the candidate solution based on its fitness. A heuristic problem-solving strategy manipulates the engineering measurements of the candidate solution to achieve a more quantitatively fit solution to the land development problem. A computer output device outputs to a user documentation illustrating the fit solution to the land development problem.