Abstract: Fuzzy logic controller for a queue manager that services a plurality of queues in an iterative fashion, each queue containing an indeterminate number of elements, the manager iteratively servicing each of the plurality of queues in accordance with a dequeuing quantity parameter, Q. The controller comprises a device for determining an amount of queues having remaining elements after each servicing iteration and outputting the amount; a device for comparing the difference between a system output representing an amount of queues having remaining elements in a current iteration with a previous output representing an amount of queues having remaining elements in a predecessor iteration, the output difference indicating a first dequeuing trend; and, a device for adjusting the dequeuing quantity parameter Q based on the dequeuing trend, the adjusting device capable of driving the system output toward a predetermined setpoint quantity after each servicing iteration.

Abstract: In a negotiation performed among a plurality of inferring units that are mutually connected through a network, at any time of which a consistent inferred state is obtained, the inferred state is accepted by the units as a settled inferred state. Since each inferring unit can perform an inference based on the settled inferred state, the inference can be effectively performed. Particularly, when a negotiation is hierarchically performed (hierarchical negotiation) or when a plurality of negotiations are performed in parallel (multi-negotiation), each negotiation can be consistently effected. In addition, in a negotiating process, each inferring unit negotiates with other inferring units using negotiation strategies having a standpoint of priority (role) and tolerance (transaction) of a request in the negotiation. In addition, each inferring unit autonomously changes the negotiation strategies for use corresponding to the negotiation history, the current internal state, and the current external state.

Abstract: A system and method for training a neural network that ceases training at or near the optimally trained point is presented. A neural network having an input layer, a hidden layer, and an output layer with each layer having one or more nodes is presented. Each node in the input layer is connected to each node in the hidden layer and each node in the hidden layer is connected to each node in the output layer. Each connection between nodes has an associated weight. All nodes in the input layer are connected to a different historical datum from the set of historical data. The neural network being operative by outputting a prediction or classification, the output of the output layer nodes, when presented with input data. The weights associated with the connections of the neural network are first adjusted by a training device. The training device then iteratively applies a training set to the neural network, the training set consisting of historical data.

Abstract: The present invention provides a method of performing statistical classification that can resolve conflict in independent sources of information, thereby creating a robust statistical classifier that has superior performance to classifiers currently available. Additionally, the present invention is automatically trainable, yielding improved classification performance. The present invention may be embodied in a method of statistically classifying events or objects, including the steps of gathering sets of information representative of features of an object or event; creating basic probability assignments based on said sets of information; determining a coarse information set from said sets of information; performing coarsening on said sets of information; performing linear superposition on each feature; and combining all features to reach a conclusion.

Abstract: For controlling a process, manipulated variables for a plurality of actuators that act on the process are calculated in a control device from measured values for output quantities of the process.To optimize the control of the process by control device (7), the actuator efficiencies (w.sub.11, . . . w.sub.nm) that describe the dependence of the changes in the output quantities (dy.sub.1, . . . , dy.sub.m) on the changes in the manipulated variables (du.sub.1, . . . , du.sub.n) for each actuator are learned in a neural network (10) and sent to the control device (7) to improve the calculation of the manipulated variables (u.sub.1, . . . , u.sub.

Abstract: An automatic aider of human cognitive functions for the operation of computerized displays. The invention estimates in real-time the mental decision made in response to a displayed stimulus from the single-even, transient, evoked cerebral potential and the corresponding visual response. The invention attains high accuracy levels of decision classification by combining a unique parametric model of the cerebral potential with advanced techniques drawn from numerical analysis, artificial intelligence, and nonlinear regression analysis. The invention uses an expert system to determine the decision aiding to be provided to the human operator from the disparities among the estimations and the decisions that are expected from task scripts for the displayed stimulus.

Abstract: An apparatus for deciding a shift pattern suitable for a driver's habit using a neural network operation and fuzzy inference and a control method thereof which perform a neural network operation by inputting a driver's driving operation quantity as a deciding condition of a shift pattern, and decide an optimal shift pattern by performing fuzzy inference from the output from the a neural network operation.

Abstract: An electronic device for performing convolution operations comprises shift registers for receiving binary input values representative of an original matrix, synapses for storing weights correlated with a mask matrix, and neurons for outputting a binary result dependent on the sum of the binary values weighted by the synapses. Each synapse has a conductance correlated with the weight stored and dependent upon the binary input value. Each neuron generates the binary result in dependence on the total conductance of the corresponding synapses.

Abstract: In a rule-based processor, populations of component agents are enabled to self-assemble into compound agents such as to give a degree of distributed artificial intelligence in a relatively simple system. The majority of component agents have top and bottom decision structures into which requesting agents can insert rules: the rule selected by a decision structure is the one which is valid following evaluation of a priority criteria and leads to the placing of a marker from the selecting agent to the requesting agent. With the two decision structures, behaviour trees may be formed from the top down or the bottom up, as required by the instantaneous situation to which the compound agent is responding.