Abstract: Disclosed is a general learning framework for computer implementation that induces sparsity on the undirected graphical model imposed on the vector of latent factors. A latent factor model SLFA is disclosed as a matrix factorization problem with a special regularization term that encourages collaborative reconstruction. Advantageously, the model may simultaneously learn the lower-dimensional representation for data and model the pairwise relationships between latent factors explicitly. An on-line learning algorithm is disclosed to make the model amenable to large-scale learning problems. Experimental results on two synthetic data and two real-world data sets demonstrate that pairwise relationships and latent factors learned by the model provide a more structured way of exploring high-dimensional data, and the learned representations achieve the state-of-the-art classification performance.

Abstract: The present invention has its objective to provide a defect classification apparatus which suppresses over-fitting and accurately classify the defect type of a defect.

Abstract: A message processing system and method that recommends actions for incoming messages based upon past historical email behavior information. The historical email behavior information represents a user's behavior for a plurality of past messages and an action is recommended based on a comparison of the incoming messages to the historical email behavior information.

Abstract: A method for assessing survivability of a distributed automation power grid due to a failure including: identifying a location of a failed section of a grid and isolating the section; aggregating sections of the grid downstream from the failed section into a first group and sections of the grid upstream from the failed section into a second group; computing, using a Markov chain, a first parameter indicating whether a communication network of the grid is operational after the failure; computing, using a Markov chain, a second parameter indicating whether enough backup energy exists for the second group after the failure; and computing, using a Markov chain, a third parameter based on the first and second parameters, wherein the third parameter is indicative of the time required for the grid to be restored to its operating state prior to the failure, and the third parameter is computed by evaluating ESAIDI ? ( ? ) = ? j = 1 C ? ? ? j ? ? ? ( M _ j ? ( ? ) D _ j ? ( ? ) )

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating hash functions. In one aspect, a method includes generating hash functions by sequentially determining a weight vector for each hash function to maximize an accuracy measure derived from current constraint weights and updating the current constraint weights for use in calculating a weight vector of a next hash function in the sequence. In another aspect, the determined weight vector maximizes an accuracy measure and a variance measure. In still another aspect, a method includes generating an adjusted covariance matrix and generating a sequence of hash functions from the adjusted covariance matrix. In still another aspect, a method includes sequentially generating a sequence of hash functions, where the weight vectors for any previously generated hash functions are used to identify constraints used to generate the weight vector for each next hash function in the sequence.

Abstract: Embodiments of the present invention can comprise systems and methods for managing filesystems and can provide a programmable filesystem with active rules and policies, an n-tier network filesystem, stack mounting, a union filesystem with write-through semantics, a filesystem middleware with selective delegation, a filesystem with a statistical warehouse and/or other management functionality.

Abstract: The present invention provides a technique for building natural language parsers by implementing a country and/or jurisdiction specific set of training data that is automatically converted during a build phase to a respective predictive model, i.e., an automated country specific natural language parser. The predictive model can be used without the training data to quantify any input address. This model may be included as part of a larger Geographic Information System (GIS) data-set or as a stand alone quantifier. The build phase may also be run on demand and the resultant predictive model kept in temporary storage for immediate use.

Abstract: Described is a system and method for training a machine learning network. The method comprises initializing at least one of nodes in a machine learning network and connections between the nodes to a predetermined strength value, wherein the nodes represent factors determining an output of the network, providing a first set of questions to a plurality of users, the first set of questions relating to at least one of the factors, receiving at least one of choices and guesstimates from the users in response to the first set of questions and adjusting the predetermined strength value as a function of the choices/guesstimates. The real and simulated examples presented demonstrate that synthetic training sets derived from expert or non-expert human guesstimates can replace or augment training data sets comprised of actual training exemplars that are too limited in size, scope, or quality to otherwise generate accurate predictions.

Abstract: A method for merging information of high semantic level representing complex situations including several objects originating from several sensors ci, said information possibly being heterogeneous, said method including acquiring the various items of information arising from said sensors in raw form and transforming them so as to cast them into the form of conceptual graphs, a conceptual graph representing several concepts and the relations which exist between them, and being composed of entity nodes and relation nodes, with E the set of concept nodes defined on a support S, G1 and G2 two conceptual graphs defined on S, wherein the method determines a merging strategy denoted strategymerge as follows: strategymerge=fmerge?fcomp:E×E?E?{E×E} and said merging strategy is applied to the various items of information taking the form of graphs.

Abstract: In accordance with one embodiment, a special purpose computer can be implemented for processing a linear optimization problem capable of being represented in the form [A][X]+[I][Y]=[B] and wherein the linear optimization problem can also be represented in the form [E][A][X]+[E][I][Y]=[E][B]. The computer may comprise a first processor; a plurality of row processors each configured to store a row of the matrix [E]; a computer memory in communication with the first processor and in communication with each row processor so that each row processor can read from the computer memory and write to the computer memory; wherein the first processor is configured to signal all of the row processors to process data related to the linear optimization problem.

Abstract: Authority over an artificial intelligence (AI) asset can be controlled among two or more processing devices running a common program over a network using a technique in which authority can be transferred. A first processing device can exercise authority over the AI asset by executing code that controls one or more actions of the AI asset according to a decision tree. The decision tree can determine whether to engage the program asset based on criteria other than a distance between the AI asset and the program asset. The first processing device can broadcast a state of the AI asset to one or more other devices running the program. If the decision tree determines that the AI asset should engage a program asset over which another processing device has authority the first processing device can relinquish authority over the AI asset and transfer authority to the other device.

Abstract: A method of optimizing a drilling operating parameter or a drilling system parameter for a drilling assembly employing at least first and second distinct cutting structures includes entering at least one design parameter for each of the cutting structures into a trained artificial neural network. At least one of the design parameters of the first cutting structure may be optionally combined with at least one of the design parameters of the second cutting structure. The combined design parameter may also be entered into the artificial neural network.

Abstract: A system for managing a user's exposure to the ultraviolet radiation including a user input interface, display circuitry including a screen, control circuitry including at least one processor wherein the control circuitry is configured to communicate with the display circuitry and the user input interface, data storage means for storing program instructions that, when implemented by the control circuitry, are configured to determine UV index information corresponding to a user's location, communicate with the user input interface to retrieve information for at least one user parameter, calculate at least one recommended exposure time based on the UV index information and the user parameter information, and communicate with the display circuitry to display the recommended exposure time.

Abstract: Boolean graphs and complex non-intuitive editors for authoring complex nested Boolean expressions can result in erroneous evaluation results by relying exclusively on the ability of an author of the Boolean expression to accurately incorporate operands into the Boolean expression to achieve the desired result. Functionality for analyzing the effect of operator changes can be incorporated into an intuitive graphical user interface for authoring the Boolean expression. The impact of the operator changes can be determined based on a position of the operator in the Boolean expression. Multiple evaluation orders for evaluating the operators and their associated operands can also be identified and presented to the programmer to resolve ambiguities, that may arise while evaluating the Boolean expression. The organization of the Boolean expression can be graphically depicted to reflect an evaluation order of the operators and their associated operands that comprise the Boolean expression.

Abstract: In particular embodiments, a method includes receiving a query for data in data sets that are within a specified range, constructing a first binary decision diagram (BDD) representing the specified range, and constructing a third BDD representing the data in the specified range by performing an AND operation between the first BDD and a second BDD representing the data sets.

Abstract: According to various embodiments of the present invention, user performance and/or motivation for a computing system may be maximized by optimizing one or more target components of a user interface of the computing system. The target components may be aspects of the user interface that is perceived by the user. One or more input features and one or more output features may be identified, and data regarding these input and output features may be gathered. This data may be compared with the results generated by a set of candidate artificial intelligence algorithms to determine which of them provides the best fit with the data collected. Then, the selected artificial intelligence algorithm may be applied to the user interface to iteratively change the target components over time until the optimal settings for each user are discovered.

Abstract: An admission control system for a cloud database includes a machine learning prediction module to estimate a predicted probability for a newly arrived query with a deadline, if admitted into the cloud database, to finish its execution before said deadline, wherein the prediction considers query characteristics and current system conditions. The system also includes a decision module applying the predicted probability to admit a query into the cloud database with a target of profit maximization with an expected profit determined using one or more service level agreements (SLAs).

Abstract: A method, apparatus, system, article of manufacture, and computer readable storage medium provide the ability to predict and utilize a user's attributes. A sample user behavior and a sample user attribute are collected. A model is trained based on the sample user behavior and sample user attribute. Using the model, a probability of a predicted user attribute based on the sample user behavior is predicted. Using the model and the probability, the predicted user attribute is fuzzily determined based on a real user behavior. The predicted user attribute is used to improve a user's experience.

Abstract: A signal analyzer is operable to receive one or more signals and then to analyze the signals with respect to particular properties before outputting an indication as to the classification of the signal with respect to each property. The indication is in the form of a plurality of belief index ratings, the magnitude of each belief index rating relating to the likelihood of the corresponding classification being accurate. This is achieved by incrementing or decrementing one or more individual belief index values in response to the identification of predetermined signal features by the signal processing means. Which belief index values are incremented or decremented and by how much is determined by the particular signal feature detected. Data is retrieved from a suitable look up table. Data storage is operable to store a copy of the received signals for future reference.

Abstract: A knowledge tracking and analysis system and method. The method may include: an artifact evaluator that receives various use parameters for each knowledge artifact, operates on the received use parameters, and generates respective parameter scores; and a knowledge index calculator that compiles parameter scores and generates a knowledge index indicative of use of the knowledge entities during the project. The system and method allow analysis and tracking of knowledge artifacts in future projects.