Abstract: A method of verifying pretrained, static, feedforward neural network mapping software using Lipschitz constants for determining bounds on output values and estimation errors is disclosed. By way of example, two cases of interest from the point of view of safety-critical software, like aircraft fuel gauging systems, are discussed. The first case is the simpler case of when neural net mapping software is trained to replace look-up table mapping software. A detailed verification procedure is provided to establish functional equivalence of the neural net and look-up table mapping functions on the entire range of inputs accepted by the look-up table mapping function. The second case is when a neural net is trained to estimate the quantity of interest form the process (such as fuel mass, for example) from redundant and noisy sensor signals.

Abstract: A synapse element consisting of a smaller number of elements utilizing common semiconductor technology, and a neuron circuit and a neuron device using the synapse elements are provided. The synapse element comprises a transistor set consisting of two MIS transistors connected in series. The first transistor adjusts the effective &bgr;-value of the transistor set so as to correspond to the weight factor &ohgr; via voltage applied to its gate electrode, and the second transistor switches the current according to input voltage to its gate electrode, so that output of the transistor set represents synapse output &ohgr;X. A voltage holding element and a switching element furnished to the gate of the first transistor give the neuron device a learning ability.

Abstract: A data processing and analysis system, and a method of using the same, for discovering implicit relationships in data. The method is executed in a computer system capable of receiving input data comprised of expert knowledge, empirical data, and user-defined constraints for any application domain. The system and method provide any pre-processing the input data may require, and perform feature selection and extraction on the input data. Further, the system and method generate a graphical representation of the implicit relationships in the input data, indicating relationships between both class variables and feature variables. Also generated is a classifier that provides a semantic and statistical justification of its classification results which further provides: statistical relevancy of the data set, including an indication of the undersampled regions of the data space; a data analysis specific to a desired level of confidence; and a sound decision theoretical foundation for classification thresholding.

Abstract: A method and means for speeding up simulation in a communication system use Importance Sampling (IS) involving fitted stochastic processes that are artificially altered and fed to the system as inputs in order to shorten the time between so called rare events. The simulation needs to be compensated for the alteration so that the simulation remains unbiased. The problem of finding optimal alternations can be reduced to find optimal bias parameters for the stochastic processes. Here an Artificial Neural Network (ANN) together with a Statistical Bias Optimizer is used in a first phase where the ANN is trained with input parameters and their corresponding optimal output bias parameters. In an application phase, only the trained ANN is used to very quickly provide optimal output samples for the final simulation.

Abstract: A robotic system which controls the movement of a surgical instrument in response to voice commands from the user. The robotic system has a computer controlled arm that holds the surgical instrument. The user provides voice commands to the computer through a microphone. The computer contains a phrase recognizer that matches the user' speech with words stored in the computer. Matched words are then processed to determine whether the user has spoken a robot command. If the user has spoken a recognized robot command the computer will move the robotic arm in accordance with the command.

Abstract: This invention provides methods and systems for classification of data instances using tree-based classifiers. Tree-based classifiers are first built according to a particular structure using a technique known as “boosting” wherein simple rules are successively added to the tree-based classifier until the unit tree-based classifier exhibits satisfactory error performance. The resulting classifier can then accurately classify an instance of data submitted to it and the classifier can optionally provide a measure of confidence for the classification result produced by the tree-based classifier.

Abstract: A method for determining state of health (SOH) of an electrochemical device using fuzzy logic (i.e., an intelligent system) is presented. State of health of an electrochemical device is determined by an internal characteristic parameter (or external operating and environmental conditions) of the electrochemical device and a characteristic parameter of a load with an intelligent system. The electrochemical device comprises such devices as primary (“throwaway”) batteries, rechargeable batteries, fuel cells, hybrid batteries containing a fuel cell electrode or electrochemical supercapacitors. The intelligent system is trained in the relationship between the characteristic parameters of the electrochemical device, the characteristic parameters of the load and the SOH of the electrochemical device.

Abstract: A method for transforming a fuzzy logic system into a neural network, where, in order to simulate membership functions, sigmoid functions are linked together in such a way that, even after the optimization of the neural network, back-transformation of the neural network into a, fuzzy logic system is possible. The advantage of the method described is that a fuzzy logic system can be transformed, in particular component by component, into a neural network and the latter can then be optimized as a whole, i.e. all the components together. The possibility of back-transforming the trained neural network ultimately means that an optimized fuzzy logic system can be obtained. This advantageously makes it possible to use, in particular, standardized fuzzy system software for describing the optimized fuzzy logic system.

Abstract: A method and apparatus are provided for multi-class, mutli-label information categorization. A weight is assigned to each information sample in a training set, the training set containing a plurality of information samples, such as text documents, and associated labels. A base hypothesis is determined to predict which labels are associated with a given information sample. The base hypothesis predicts whether or not each label is associated with information sample or predicts the likelihood that each label is associated with the information sample. In the case of a document, the base hypothesis evaluates words in each document to determine one or more words that predict the associated labels. When a base hypothesis is determined, the weight assigned to each information sample in the training set is modified based on the base hypothesis predictions.

Abstract: A method of operating a target system wherein small changes in input variables produce small changes in output variables in a manner permits system learning on a time dependent basis. The rate of change of the system output is directly dependent upon the product of the rate of change of the system input and a matrix consisting of input variables and antecedent and consequent parameters. A preferred performance criterion is obtained through the approximation of the said matrix to a weighted-augmented pseudo-inverse Jacobian. Off-line, the system undergoes a series of iterations using a wide range of input variables wherein the actual outputs are compared to the desired outputs and optimized values for the antecedent and consequent parameters are obtained and passed back to the said matrix for use in the subsequent iteration.

Abstract: This invention is directed to a method for predicting the flammability of complex mixtures using critical variables of structural groups comprising training data from the critical variables of each structural groups, the critical variables comprising compositional and thermochemical data from each of the structural groups to produce a neural network model; testing the trained data from the neural network model; and validating the trained and tested data from the neural network to accurately predict the flammability limit of an analogous complex mixture having similar structural groups.

Abstract: An adjustment control method and apparatus of this invention assume that a dependency relationship table representing qualitative characteristics, in which manipulated variables are classified in units of change patterns of controlled variables, and are characterized in that an adjustment rule for adjustment is generated. An adjustment control method and apparatus of this invention is characterized in that it is determined whether the current object situation exhibits an exceptional behavior (vibration/saturation), on the basis of selection of an adjusted controlled variable and a manipulated variable obtained from an automatically generated adjustment rule and a past operation in response to an occasionally output deviation. If it is determined that no exceptional behavior is observed, the instruction of the generated adjustment rule is performed, otherwise, the correction amount of the.

Abstract: An apparatus and system are provided for retrieving electronic records from a database of records, preferably in the form of medical records. The apparatus utilizes a knowledge base in the form of a data structure set forth in a description language. The knowledge base includes representations of a plurality of concepts hierarchically linked with other concepts within a plurality of classes, and non-hierarchically linked with concepts in other classes. A query manager receives a user query and translates the query into the description language. A records analyzer is coupled to the query manager, database and knowledge base. The records analyzer analyzes the records in the database based on the user query and the classes and the hierarchical and non-hierarchical relationships in the knowledge base.

Abstract: An acoustic imaging system for forming acoustic beams approximating an optimum acoustic beam for the directional transmission or reception of acoustic energy. Maximum and minimum dependent beamform factors are determined from initial beamform factors and an initial parent population of chromosomes is generated, each chromosome including a gene corresponding to a dependent beamform factor and representing an initial candidate beam and subsequent parent populations are generated by cloning of the surviving populations. A child population is generated by exchanging statistically selected pairs of genes of the parent population and generating a mutated population. A surviving population is selected from the mutated population of the mutated population with a fitness criteria. When a chromosome of the surviving population meets the solution criteria, the genes of the surviving population having the best match to the fitness criteria are selected to forming a beam.

Abstract: An engine diagnostic system includes a bit-serial based recurrent neuroprocessor for processing data from an internal combustion engine in order to diagnose misfires in real-time and reduces the number of neurons required to perform the task by time multiplexing groups of neurons from a candidate pool of neurons to achieve the successive hidden layers of the recurrent network topology.

Abstract: A rules based decision management system using online analytical processing (OLAP) technology for dynamic assessment of strategy results. Generally, a rules based decision management system applies strategies which produce results. The results are aggregated over time, typically in accordance with values of a discrete dimension and ranges of a continuous dimension, to prepare for the application of OLAP technology. Date stamping can be used so that, when aggregating the results, different values and different ranges can be valid for different periods of time. OLAP technology is then applied to the aggregated results, to evaluate the applied strategies.

Abstract: A fuzzy logic controller system comprising a plurality of fuzzy logic controller units cascadably coupled to one another, each controller having a membership processor for receiving a single input signal corresponding to a sensed parameter from a controlled system and performing fuzzy inference operations based on the input signal for outputting an incremental output signal indicative of a requested change in a controlled output from a current level, a permissions processor for scaling the incremental output signal in response to a fulfillment signal from another controller, a processor responsive to the membership processor and to the permissions processor for providing a fulfillment signal to another controller indicative of a permissible range of freedom for scaling the incremental output signal of the another controller in the cascading sequence, an adder for summing each of the incremental output signals from the plurality of controllers and applying the sum to the controlled output for providing a corre

Abstract: A system and method for enhancing knowledge discovery from data using multiple learning machines in general and multiple support vector machines in particular. Training data for a learning machine is pre-processed in order to add meaning thereto. Pre-processing data involves transforming the data points and/or expanding the data points. By adding meaning to the data, the learning machine is provided with a greater amount of information for processing. With regard to support vector machines in particular, the greater the amount of information that is processed, the better generalizations about the derived data. Multiple support vector machines, each comprising distinct kernels, are trained with the pre-processed training data and are tested with test data that is pre-processed in the same manner. The test outputs from multiple support vector machines are compared in order to determine which of the test outputs if any represents a optimal solution.

Abstract: An intelligent agent workbench is disclosed for developing a computer-based intelligent agent that can be inserted between a system and an individual performing a pre-defined function with the system and/or that can supplant such function performed by such individual. In the workbench, an agent author allows a workbench user to create and modify an intelligent agent in a graphical representation. The workbench also has an execution engine for executing a non-graphical representation of the agent both within the workbench and outside the workbench as a stand-alone agent, and an interpreter for automatically translating between the non-graphical and graphical representations. The agent has an instance structure representing a current problem instance, perception structures, task performance structures, and a priority structure.

Abstract: The invention relates to a method of coding and storing fuzzy logic rules, and to a circuit architecture for processing such rules. The method provides for at least one inference rule of the IF/THEN type, having a predetermined number of antecedent parts of fuzzy variables and at least one consequent part, to be dismembered and stored into memory words to allow subsequent processing using logic operators of the AND/OR/NOT type. The coding of rules and variables is effected sequentially. Thus, the occupation of memory locations can be minimized. Specifically, the rules are coded through a multi-word description, such that the number of words coding each rule is a varying number dependent on the number of antecedent parts in the rule.