Abstract: A method for predicting future responses from large sets of dyadic data includes measuring a dyadic response variable associated with a dyad from two different sets of data; measuring a vector of covariates that captures the characteristics of the dyad; determining one or more latent, unmeasured characteristics that are not determined by the vector of covariates and which induce local structures in a dyadic space defined by the two different sets of data; and modeling a predictive response of the measurements as a function of both the vector of covariates and the one or more latent characteristics, wherein modeling includes employing a combination of regression and matrix co-clustering techniques, and wherein the one or more latent characteristics provide a smoothing effect to the function that produces a more accurate and interpretable predictive model of the dyadic space that predicts future dyadic interaction based on the two different sets of data.

Abstract: A knowledge model discovery system is configured to provide an interactive view having simultaneously displayed sub-views that include a relational data element view and a geophysical view. Using the interactive view, a user may perform search and analysis of information organized with a knowledge management tool in conjunction with geographic information. The relational data element view may provide a relational visualization that displays data elements provided with the knowledge management tool as narrowed by the context of a user analysis. The geophysical view may provide a geographically based depiction of the same data elements using a similar context. The relational data element view and the geophysical view may be operated in coordination to maintain cohesiveness and similar context of the information displayed in the views.

Abstract: A system and method that optimizes reduce operations by consolidating the operation into a limited number of participating processes and then distributing the results back to all processes to optimize large message global reduce operations on non power-of-two processes. The method divides a group of processes into subgroups, performs paired exchange and local reduce operations at some of the processes to obtain half vectors of partial reduce results, consolidates partial reduce results into a set of regaining processes, performs successive recursive halving and recursive doubling at a set of remaining processes until each process in the set of remaining process has a half vector of the complete result, and provides a full complete result at every process.

Abstract: A performance evaluating apparatus for a rule is provided, which is capable of evaluating a business value of a rule applied to the operation and management of an information system through autonomous control. The performance evaluating apparatus for a rule collects, respectively be fore activation processing that is prescribed by a rule and after completion of the execution of the processing, a rule execution history and a system state information indicating a state of a monitored information system that is necessary for calculation of the business value. The business value is calculated from the collected system state information and rule execution history by a given calculation formula.

Abstract: A learning apparatus includes a storage unit configured to store a network formed by a plurality of nodes each holding dynamics; a learning unit configured to learn the dynamics of the network in a self-organizing manner on the basis of observed time-series data; a winner-node determiner configured to determine a winner node, the winner node being a node having dynamics that best match the time-series data; and a weight determiner configured to determine learning weights for the dynamics held by the individual nodes according to distances of the individual nodes from the winner node. The learning unit is configured to learn the dynamics of the network in a self-organizing manner by degrees corresponding to the learning weights.

Abstract: This invention generally relates to methods, apparatus and computer program code processing digital data using non-negative matrix factorisation.

Abstract: A tensor voting scheme which can be used in an arbitrary number N of dimensions, up to several hundreds. The voting scheme can operate on unorganized point inputs, which can be oriented or unoriented, and estimate the intrinsic dimensionality at each point. Moreover it can estimate the tangent and normal space of a manifold passing through each point based solely on local operations.

Abstract: A method for moving an agent with a compiled rule set from a first execution environment to a second execution environment, comprising initiating the agent move, sending the agent from the first execution environment to the second execution environment and receiving the agent at the second execution environment.

Abstract: Automated assistance is provided for managing rules and/or guidelines regarding the usage of branded content for a project, e.g., an advertising campaign, event, signage, decoration, and the like. One or more queries regarding the project are presented to the user. Based on the user's responses to the queries, branded content that is approved for the project's intended use(s) is automatically identified and made available to the user. The determination of which digitized content (media objects) to provide to the user is based at least in part on brand rules and/or guidelines that can be relatively static or dynamically updated by the current user.

Abstract: After initial clusters having only one component are formed, a conditional probability P(Ci|C?k) is determined for the cluster Ci being included in an order on condition that cluster C?k is included in the order. If P(Ci|C?k) is greater than a first threshold value S1, a new cluster Cn having all the components of clusters Ci, C?k is formed and the operations are repeated until no new clusters are formed.

Abstract: A collection of web pages is considered as a directed graph in which the pages themselves are nodes and the hyperlinks between the pages are directed edges in the graph. A trusted entity identifies training examples for spam pages and normal pages. A random walk is conducted through the directed graph that includes the collection of web pages and the stationary probabilities, and transitional probabilities, among the nodes in the directed graph are obtained. A classifier training component estimates a classification function that changes slowly on densely connected subgraphs within the directed graph. The classification function assigns a value to each of the nodes in the directed graph and identifies them as spam or normal pages based upon whether the value meets a given function threshold value.

Abstract: The present invention relates to a system for managing a plurality of multi-field classification rules. The system provides a first table that includes a plurality of entries corresponding to a plurality of rules relating to an ingress context and a second table that includes a plurality of entries corresponding to a plurality of rules relating to an egress context. The system also includes a network processor for classifying packets of information, wherein the network processor is programmed to utilize the first table and the second table to identify any rules relating to the ingress context and any one rules relating to the egress context that match a search key.

Abstract: The present invention relates to a method and computer system device for applying a plurality of rules to data packets within a network computer system. A filter rule decision tree is updated by adding or deleting a rule. If deleting a filter rule then the decision tree is provided to a network data plane processor with an incremental delete of the filter rule. If adding a filter rule then either providing an incremental insertion of the filter rule to the decision tree or rebuilding the first decision tree into a second decision tree responsive to comparing a parameter to a threshold. In one embodiment the parameter and thresholds relate to depth values of the tree filter rule chained branches. In another the parameter and thresholds relate to a total count of rule additions since a building of the relevant tree.

Abstract: Described are techniques for using statistical analysis to reduce the number of samples required in accordance with statistical analysis confidence intervals to verify correctness of a component. These techniques may be used in verification of a neural network or other hardware or software component.

Abstract: Monitoring dynamic units that operate in complex, dynamic environments, is provided in order to classify and track unit behavior over time. When domain knowledge is available, feature-based models may be used to capture the essential state information of the units. When domain knowledge is not available, raw data is relied upon to perform this task. By analyzing logs of event messages (without having access to their data dictionary), embodiments allow the identification of anomalies (novelties). Specifically, a Normalized Compression Distance (such as one based on Kolmogorov Complexity) may be applied to logs of event messages. By analyzing the similarity and differences of the event message logs, units are identified that did not experience any abnormality (and locate regions of normal operations) and units that departed from such regions.

Abstract: A decision tree is provided as a machine learning classifier to predict a user attribute, such as a geographical location of a user, based on a network address. More specifically, the decision tree is constructed via machine learning on a set of sample data that reflects a relationship between a network address and a user attribute of a “known user” whose profile information is recognizable. For a given network address, the decision tree can be used as a machine learning classifier to predict the most likely user attribute of a potential user. With the predicted attribute, a network service can target a group of potential users for various campaigns without recognizing the identities of the potential users.

Abstract: A system and method for processing documents by utilizing the textual content and layout of the documents, including visual indicators, to more efficiently and reliably process the documents across various document types. The system and method identifies visually distinguishable elements within the document, such as section and sub-section boundary indicators, to mark, divide and label the boundaries and content type such that the sections are more clearly identifiable and easily processed. The system and method uses known elements, including section heading types, keywords, section type classifiers, sub-section heading constructs, stop words, and the like to adaptively identify and process a broad range of document types. The system and method continually refines and updates these known elements and allows users to discover and define new elements for further refinement and updating.

Abstract: The invention is a technique for performing sampling in connection with Markov Chain Monte Carlo simulations in which no attempt is made to limit the selected samples to a selected slice of the entire sample domain, as is typical in Markov Chain Monte Carlo sampling. Rather, samples are taken from the entire domain and any samples that fall below a randomly selected probability density level are discarded.

Abstract: A method for the automatic sequencing of the specifications of a computer comprises an analysis of the specifications constituted by nodes with a simplification and distribution of these nodes into at least two types of nodes, and an assigning of these nodes in repetitive sub-cycles of processing tasks in order to obtain a substantially uniform distribution of the tasks.

Abstract: The invention relates to a method for determining a time for retraining a data mining model, including the steps of: calculating multivariate statistics of a training model during a training phase; storing the multivariate statistics in the data mining model; evaluating reliability of the data mining model based on the multivariate statistics and at least one distribution parameter, and deciding to retrain the data mining model based on an arbitrary measure of one or more statistical parameters including an F-test statistical analysis.