Abstract: A computer assisted/implemented method for developing a classifier for classifying communications includes roughly four stages, where these stages are designed to be iterative: (1) a stage defining where and how to harvest messages (i.e., from Internet message boards, ews groups and the like), which also defines an expected domain of application for the lassifier; (2) a guided question/answering stage for the computerized tool to elicit the user's criteria for determining whether a message is relevant or irrelevant; (3) a labeling stage where the user examines carefully-selected messages and provides feedback about whether or not it is relevant and sometimes also what elements of the criteria were used to make the decision; and (4) a performance evaluation stage where parameters of the classifier training are optimized, the best classifier is produced, and known performance bounds are calculated.

Abstract: A method for predicting a future occurrence of an event involves obtaining a history of prior occurrences of the event. A plurality of variables is created that are associated with the event. Weights are assigned to each variable. An artificial neural network is accessed and trained with the history of past occurrences of the event by comparing an output of the artificial neural network to the past occurrence of the event. The weights are adjusted until the output corresponds to the past occurrence of the event.

Abstract: A method of implicitly forming a context in a dialogue with a computer includes forming megacategories linking one or more attributes to one of a plurality of response templates. The method also includes determining a megacategory by comparing attributes having true variables from a previous input to attributes having true variables in the present input and selecting an appropriate response template corresponding to the megacategory.

Abstract: A probabilistic boosting tree framework for computing two-class and multi-class discriminative models is disclosed. In the learning stage, the probabilistic boosting tree (PBT) automatically constructs a tree in which each node combines a number of weak classifiers (e.g., evidence, knowledge) into a strong classifier or conditional posterior probability. The PBT approaches the target posterior distribution by data augmentation (e.g., tree expansion) through a divide-and-conquer strategy. In the testing stage, the conditional probability is computed at each tree node based on the learned classifier which guides the probability propagation in its sub-trees. The top node of the tree therefore outputs the overall posterior probability by integrating the probabilities gathered from its sub-trees. In the training stage, a tree is recursively constructed in which each tree node is a strong classifier. The input training set is divided into two new sets, left and right ones, according to the learned classifier.

Abstract: An expert system is used to recommend a customized solution for a customer. The expert system includes a set of scenarios each of which includes default facts and is associated with a detailed rule base that when applied in isolation to the default facts fully determines a prototype solution. A particular scenario is selected from the set based at least on user input. A customized solution is generated by applying the detailed rule base associated with the particular scenario to facts based at least on the user input and any default facts of the particular scenario that complement and do not conflict with the facts that are based at least on the user input. The customized solution is iteratively refined.

Abstract: The present invention provides a unique method and user interface that facilitates accessing and browsing objects in which a user begins with a center object (e.g., one or a few focal objects) displayed on a screen and related objects are populated on the screen as well. The related objects can be further organized into clusters whereby each cluster or grouping of objects expands on a particular attribute of the center object. The attributes correspond to metadata. Thus, the objects are populated based upon the metadata of the center object. According to one aspect, the user can access one or more specific objects having a plurality of attributes and then relax at least one of the attributes to see what other objects share at least one attribute with the center object. According to another aspect, the object having the closest match to a search request can be centrally displayed with other close matches arranged by their respective metadata.

Abstract: An embodiment includes a prediction system for initiating a data transfer to a decision system. The prediction system is configured to identify a decision, the decision being a result of a computation of the decision system according to a set of predefined rules and input data. The prediction system is further configured to identify predicted input data representing a portion of the input data and to initiate a transfer of the predicted input data to the decision system prior to the computation of the decision.

Abstract: Methods and apparatus are provided for determining refrigerant charge in a vapor compressor system (VCS) of an aircraft. The methods and apparatus comprise the following steps of, and/or means for, generating a data set from historical data representative of a plurality of VCS operating conditions over time, identifying one or more steady-state data points in the generated data set, forming a revised data set that includes at least the steady-state data points, using principal components analysis (PCA) to derive values for a plurality of minimally correlated input variables, supplying the derived values for the plurality of minimally correlated input variables and the corresponding values for the VCS refrigerant charge in the revised data set to a nonlinear neural network model, and deriving a simulator model characterizing a relationship between the plurality of minimally correlated input variables and the VCS refrigerant charge.

Abstract: A fault identification system analyzes alarm information and the associated contextual information to identify the occurred faults. The contextual information is based on usecases, transactions, and functions associated with the electronic system under consideration. Multiple detectors that form part of knowledge repositories are used in the process of fault identification.

Abstract: In a hierarchical neural network having a module structure, learning necessary for detection of a new feature class is executed by a processing module which has not finished learning yet and includes a plurality of neurons which should learn an unlearned feature class and have an undetermined receptor field structure by presenting a predetermined pattern to a data input layer. Thus, a feature class necessary for subject recognition can be learned automatically and efficiently.

Abstract: This invention provides a means to minimize the costs of technical and business processes. These processes are comprised of resources and tasks requiring resources. The optimization consists of the best assignment of resources to tasks to minimize the costs. In the resource assignment optimization method disclosed herein, Genetically Adapted Search Agents (GASA) are employed to improve a population of possible assignments, each represented by a single variable length chromosome, where the chromosome upon which the GASA operates is a direct encoding of possible resource to task assignments and order. To manage the enlarged search space, this method uses the GASA with substring crossover to evolve the population towards better solutions. The assignments generated by this method satisfy all constraints.

Abstract: An active semiotic system creates implicit symbols and their alphabets from features, structural combination of features, objects and patters, creates models with explicit structures that are labeled with the implicit symbols, and derive other models in the same format via diagrammatic- and graph transformations. The active semiotic system treats vision as a part of a larger system that converts visual information into special knowledge structures that drive a vision process, resolve ambiguity and uncertainty via feedback projections, and provide image understanding that is an interpretation of visual information in terms of corresponding knowledge models. Mechanisms of image understanding, including mid- and high-level vision, are presented as methods and algorithms of the active semiotic system, where they are special kinds of diagrammatic and graph transformations.

Abstract: The present invention provides a unique system and method that can employ machine learning techniques to automatically fill one or more fields across a diverse array of web forms. In particular, one or more instrumented tools can collect input or entries of form fields. Machine learning can be used to learn what data corresponds to which fields or types of fields. The input can be sent to a central repository where other databases can be aggregated as well. This input can be provided to a machine learning system to learn how to predict the desired outputs. Alternatively or in addition, learning can be performed in part by observing entries and then adapting the autofill component accordingly. Furthermore, a number of features of database fields as well as constraints can be employed to facilitate assignments of database entries to form values—particularly when the web form has never been seen before by the autofill system.

Abstract: Associative memories include associative memory cells. A respective cell includes a sensor input, a prior association representation, a next association representation and an associative output. The cells are serially interconnected to form a linear array, such that the sensor inputs, the prior association representations and the next association representations of the serially connected cells are arranged in a sequence from distal to proximal cells based on affinities of associations among the series of sensor inputs. A respective cell also includes processing logic. The processing logic is responsive to the associated sensor input being active, to send a measure of the next association representation to an adjacent proximal cell and/or to send a measure of prior association representation to an adjacent distal cell.

Abstract: Dynamically filtering and classifying messages, as good messages, bulk periodicals, or spam. A regular expression recognizer, and pre-trained neural networks. The neural networks distinguish “likely good” from “likely spam,” and also operate at a more discriminating level to distinguish among the three categories above. A dynamic whitelist and blacklist; sending addresses are collected when the number of their messages indicates the sender is good or a spammer. A dynamically selected set of regular expressions input to the neural networks.

Abstract: A system in which researchers can freely and effectively use worldwide bioinformatics analysis programs available on the Internet is provided. When a bioinformatics analysis program available on the Internet is used by a user computer, a broker program is used. The broker program has a function of absorbing differences in input/output format between analysis programs, and each analysis program is provided with the broker program. A broker program-providing server stores various broker programs provided by users and makes them available to the public. When the user uses bioinformatics analysis programs available on the Internet, the user can use broker programs that are made available by the broker program-providing server and that are created by other users.

Abstract: One embodiment of the present invention provides a method for gathering information from an environment. In a first step, visual information is gathered from the environment. In a second step, information actively transmitted by objects in the environment is received. According to one embodiment, the information actively transmitted by objects in the environment is received wirelessly. In a third step, the visual information is combined with the received information in order to recognize objects.

Abstract: A system and method for classifying text includes a pre-processor, a knowledge base, and a statistical engine. The pre-processor identifies concepts in the text and creates a structured text object that contains the concepts. The structured text object is then passed to a statistical engine, which applies statistical information provided in nodes of a knowledge base to the structured text object in order to calculate a set of match scores, each match score representing the relevance of the text to an associated one of a plurality of predefined categories. The pre-processor may be implemented in the form of an interpreter which selects and executes a script that includes language- and scenario-specific instructions for performing linguistic and semantic analysis of the text.

Abstract: Embodiments of the present invention provide for a computer program of a dialog player. An artificial psychology dialog player is a software program that picks a sentence line from a repertoire according to probabilistic rules and artificial personality states. The personality for the purpose of this invention optimally has four motivational dimensions, one of which is paired with the remaining three; each pair is identified with a stage of human age. The user of the dialog player selects or identifies an age value; the software uses this value in order to determine a set of probabilities of activation for the three stages. The simulation of aging is implemented by automatically activating one of the stages according to these probabilities whenever the personality is prompted for a sentence.

Abstract: The invention provides, in one aspect, a method of finding a rule in a rules base that includes a plurality of rule entries, each having a rule and an associated objective and one or more associated circumstances. A rule finder accesses the rules base in response to a request, with a specified objective and with specified circumstances, by searching the rules base for a matching entry. Where the search fails, the rule finder searches one or more translation tables for entries having first circumstances matching those specified in the request that resulted in the failure. If found, the rule finder repeats the search of the rules base with the specified objective and with a second (or “map-to”) circumstances from matching entries in the translation tables. The rule finder can return, e.g., to an application program that issued the request, a rule found during that repeat search of the rules base.