Abstract: A neural system is disclosed for processing an exogenous input process to produce a good outward output process with respect to a performance criterion, even if the range of one or both of these processes is necessarily large and/or keeps necessarily expanding during the operation of the neural system. The disclosed neural system comprises a recurrent neural network (RNN) and at least one range extender or reducer, each of which is a dynamic transformer. A range reducer transforms dynamically at least one component of the exogenous input process into inputs to at least one input neuron of said RNN. A range extender transforms dynamically outputs of at least one output neuron of said RNN into at least one component of the outward output process. There are many types of range extender and reducer, which have different degrees of effectiveness and computational costs.

Abstract: Active vibration control (AVC) systems without online path modeling and controller adjustment are provided that are able to adapt to an uncertain operating environment. The controller (250, 280, 315, 252, 282, 317, 254, 319) of such an AVC system is an adaptive recursive neural network whose weights are determined in an offline training and are held fixed online during the operation of the system. AVC feedforward, feedback, and feedforward-feedback systems in accordance with the present invention are described. An AVC feedforward system has no error sensor and an AVC feedback system has no reference sensor. All sensor outputs of an AVC system are processed by the controller for generating control signals to drive at least one secondary source (240). While an error sensor (480, 481) must be a vibrational sensor, a reference sensor (230, 270, 295, 305, 330) may be either a vibrational or nonvibrational sensor.

Abstract: A framework for use with object-oriented programming systems provides an expert system shell that executes with a rule set developed by a framework user. The framework includes a condition manager object that keeps track of what conditions are true at any time and a rules compiler that processes the rules so they can be used by the condition manager. The conditions are a set of object classes that are organized into an inheritance hierarchy for use by the condition manager. A detector determines when to set conditions to true. When a condition state is changed, the condition manager realizes that one or more rules might indicate a particular action might now be appropriate. Therefore, the condition manager checks the rules against the conditions each time a condition changes. The expert system developer provides the set of rules and conditions appropriate for the knowledge domain of the expert system under development.

Abstract: A method of processing data such as alarms from a communications network, the network comprising a plurality of network entities, having predetermined states of operation, the method comprising the step of creating an object (88) associated with a given state of one of the entities, such as a fault state. The object comprises knowledge based reasoning capability (89) such as rules for determining whether the entity is in the given state, and the method further comprises the steps of: passing data about the network, such as alarms and events, to the object; and inferring whether the entity is in the given fault state using the rules. This enables faster correlation of alarms and simpler development and maintenance of the rules.

Abstract: A “Rapid Learner Client Service” (RLCS) system that allows a large number of end-users to obtain the benefits of a sophisticated neural-network forecasting system. Rather than purchasing or developing a forecasting system of their own, RLCS clients subscribe to a forecasting service performed by forecasting equipment located at a remote site. This allows a single highly sophisticated forecasting system to meet the forecasting needs of a large number of subscribers. This forecasting service is performed by an RLCS server that periodically and automatically accesses the subscriber's computer to obtain a fresh set of input data. Alternatively, the subscriber's computer may contact the RLCS server to initiate the process. This input data is then downloaded to the RLCS server, where it is checked and corrected for errors by imputing values for missing or deviant input values.

Abstract: A new model of motivational behavior, described as a ten-level metaperspectival hierarchy of ethical terms, serves as the foundation for an ethical simulation of artificial intelligence. This AI system is organized as a tandem, nested expert system, composed of a primary affective language analyzer, overseen by a master control unit-expert system (coordinating the motivational interchanges over real time). The systematic organization of the ethical hierarchy allows for extreme efficiency in the programming of the respective knowledge bases, employing the principles of inheritance for modeling the more abstract levels of the hierarchy: allowing a precise determination of the motivational level at issue during a given verbal interchange (defined as the passive-monitoring mode). An optional active monitoring mode permits the posing of simple yes-or-no questions, allowing for clarification of ambiguous language input.

Abstract: An agent presentation apparatus for transmitting and receiving a message and operating based on a knowledge base and a rule base. In particular, where an agent for carrying out a user request uses knowledge of a knowledge description or an action description between agents and executes communication, for example. An agent can request a WWW page locally or remotely over a network to retrieve already existing contents and make a presentation that might include one or more of images, voice, and text along with the WWW page. The apparatus can also respond to user questions about the WWW page using the knowledge description and/or the action description, for example.

Abstract: An information processing system which understands the input information as a situation, infers the requirements that situation merits and the preferred responses to those requirements using a knowledge base which has knowledge about the situations, requirements and their responses and the relations among them, and by generating these inferences further understands the input and the situation, which enables it to perform efficiently and effectively, given small amount of input information. If input information is in natural language, the linguistic knowledge base is used to analyze the input natural language information. Further, a knowledge of the user, the operations performed by the user and knowledge of the relationship between the operations of the user and the attribute of the user may also be used by the system to appropriately operate for each user.

Abstract: The invention uses fuzzy logic and/or probability distributions to automatically calculate and display the effects of contextual information on the confidence that an object in an image is an object of interest. The goal is to assist in determining the location and type of target objects of interest in that imagery. The imagery can come from any kind of imaging sensor or can be non-sensor imagery (e.g., two-and three-dimensional maps), and can be live or archived imagery. The locations of context objects can be provided by a human or a computer. The resulting set of data, including the original imagery, the locations of context objects, any results from AOD, and predictions about target object type and location, can be combined into a display that helps a human better understand where target object appears in the imagery.

Abstract: An object to be appraised is input and divided into structural elements which are stored in a storage unit. Information is input which indicates an appraisal for the object to be appraised and a combination of elements constituting the object to be appraised is corresponded to the appraisal. The correspondence is classified into criteria which are stored and an arbitrary criterion is selected. A combination of structural elements is obtained on the basis of the selected criterion and appraisal information is inputted for the obtained combination of structural elements.

Abstract: A method uses a hybrid neural network including a self organizing mapping neural network (SOM NN) and a, back-propagation neural network (BP NN) for color identification. In the method the red, green and blue (RGB) of color samples are input as features of training samples and are automatically classified by way of SOM NN. Afterwards, the outcomes of SOM NN are respectively delivered to various BP NN for further learning; and the map relationship of the input and the output defines the X,Y, Z corresponding the x, y and z values of a coordinate system of the standard color samples of RGB and IT8. By way of the above learning structure, a non-linear model of color identification can be set up. After color samples are self organized and classified by SOM NN network, data can be categorized in clusters as a result of characteristic difference thereof.

Abstract: Method and apparatus for designing a synthesis route for a target molecule is provided. The method for designing a synthesis route for a target molecule comprises: generating a plurality of individuals, wherein each individual encodes a synthesis route; decoding each individual to produce a synthesis route comprising at least one reactant molecules and at least one reaction; and determining how well the synthesis route satisfies a design goal. A computer readable medium containing instructions for a computer program executable by the computer to perform a method for designing a synthesis route for a target molecule is also provided. The apparatus comprises a parallel computer system for executing instructions of a computer program to perform a method for designing a synthesis route for a target molecule.

Abstract: Given an input sequence of data, a motif is a repeating pattern. The data could be a sequence of characters or sets of characters or even real values. In the first two cases, the number of motifs could potentially be exponential in the size of the input sequence and in the third case there could be uncountably infinite number of motifs. By suitably defining the notion of maximality and redundancy for any sequence with n characters, there exists only a linear (or no more than 3n) number of special motifs and every other motif can be generated from these irredundant motifs.

Abstract: A process and apparatus for solving the product y=Rw, where R is a matrix and w is a vector. The process includes a steps of using a matrix outer product structure of R to determine all of the unique entries in R and storing the unique monomials. A different unique number is assigned to unique entries so that each unique entry has an associated number, and the associated numbers are stored. Rw is then solved using the stored associated numbers to obtain a result in terms of the associated numbers, and converting the result to entries from the matrix R. In the preferred embodiment, the process is used for iterative training in a classification system and especially a classification system on a portable platform.

Abstract: Power industry boiler tube failures are a major cause of utility forced outages in the United States, with approximately 41,000 tube failures occurring every year at a cost of $5 billion a year. Accordingly, early tube leak detection and isolation is highly desirable. Early detection allows scheduling of a repair rather than suffering a forced outage, and significantly increases the chance of preventing damage to adjacent tubes. The instant detection scheme starts with identification of boiler tube leak process variables which are divided into universal sensitive variables, local leak sensitive variables, group leak sensitive variables, and subgroup leak sensitive variables, and which may be automatically be obtained using a data driven approach and a leak sensitivity function. One embodiment uses artificial neural networks (ANN) to learn the map between appropriate leak sensitive variables and the leak behavior.

Abstract: In a facilitating agent apparatus, a message processing section and an element processing section replace knowledge transmitted from another agent apparatus by a superior concept, combine the knowledge with another knowledge, transform ontology, or the like, based on meta-knowledge accumulated in a meta-knowledge storing section, and transmit the resulting knowledge to still another agent apparatus. Furthermore, if transmitted knowledge is denied, the message processing section and the element processing section create a message for denying their own knowledge transmitted to another agent apparatus, based on the denied knowledge.

Abstract: A method and apparatus for prediction of system reliability is disclosed. The method comprises the steps of: (a) identifying the minimal path set of components which must function for the system to function; (b) constructing a minimal path set matrix by representing the minimal path sets as binary numbers in the matrix; (c) constructing a design matrix from OR operations on sets of columns of the minimal path set matrix whose results are appended to the original minimal path set matrix; (d) constructing a vector of ones having signs related to the position in the vector; and (e) calculating the system reliability from the design matrix, vector of ones and the reliabilities of each of the components of the system. The method of the present invention also determines the structure function of the system from the design matrix, vector of ones, and the states of the components of the system. The apparatus for performing the method of the present invention comprises a programmable processor.

Abstract: A method for detecting fires and reacting thereto. A stationary, earth orbit satellite, drone aircrafts or piloted aircraft contains infrared detectors and optical means for detecting fires. A computer in the satellite, drone or piloted aircraft and/or on the ground analyzes image signals of the areas being monitored and, upon detecting infrared radiation of varying intensity and shape, indicative that a fire has started, generates signals which are indicative of the coordinate locations of the fire, the extent of the fire, the shape of the area(s) burning, the direction of movement of the fires, the speed(s) of the flame fronts, smoke condition, intensity of the fire, fire ball location(s), etc. In one form, a camera generates image signals which are digitized and recorded. Expert system logic is used to prioritize dangerous areas to assist in directing fire fighting.

Abstract: Subsearch, where a satisfiability algorithm searches through the original theory for ground clauses that satisfy some numeric property, is represented in terms of S(C,P,u,s), the set of ground instances of C that have u literals unvalued by P and s literals satisfied by the assignments in P. This representation allows an intelligent search to be performed to answer subsearch problems posed in terms of S(C,P,u,s). Intelligent Subsearch uses truth value assignments to atoms to eliminate sets of bindings to universally quantified variables within a quantified clausal constraint; the bindings being eliminated because the bindings cannot satisfy a specific statement. Intelligent subsearch backtracks away from poor choices in the search for bindings to variables within the quantified clauses. In typical uses, intelligent subsearch can reduce the time of the checking problem from O(DU) to O(D&agr;U) for some &agr;<1.

Abstract: A pattern lattice data space as a framework for analyzing data, in which both schema-based and statistical analysis are accommodated, is defined. Ways to manage the size of the lattice structures in the pattern lattice data space are described. Utilities to classify or cluster, search (find similar data), or relate data using lattice fragments in the pattern lattice data space are also described. Superpattern cone or lattice generation function, which may be used by the classification and clustering functions, is also described. In addition, a subpattern cone or lattice generation process, which may be used by the search (find similar data) and data relating functions, is also described. Finally, a function to label, in readily understandable “pidgin”, categories which classify information, is also described.