Abstract: A signal analysis device for determining the pitch of a plucked string in which propagation times of plucking transients are evaluated to determine pitch. A neural network may be employed to perform evaluation based upon groups of pulses.

Abstract: A fuzzy operation apparatus which can supply various fuzzy operators by combining basic operators provided by hardware with microprograms, performs fuzzy operations in parallel by connecting a plurality of fuzzy operation portions using a network, and performs parallel defuzzification of the result of fuzzy operations stored in the fuzzy operation portions throughout the network. The present invention establishes a network between fuzzy operation portions, performs fuzzy operation in parallel at a plurality of fuzzy operation portions, and defuzzifies each element of a fuzzy set dispersed and stored in the fuzzy operation portions such that defuzzification speed is fast. Also, since there is no need for an exclusive defuzzification processor, the cost of the system is low.

Abstract: Hyperlinear chromosomes are arrays of parameters stored in a computer readable memory, for use in implementing a genetic algorithm. Each chromosome may represent the mapping of a problem, including physical parameters, onto constituent genes which are addressed in memory as multiple, intersecting vectors in n-dimensions, where n is >2. A computing system is adapted to perform hyperlinear crossover, reproduction and fitness evaluation on the hyperlinear chromosomes. An adaptive computing system produces optimized control signals based on the attainment of a desired level of fitness of one or more hyperlinear chromosomes.

Abstract: An analysis protocol embodied in an expert system is articulated into portions to provide optimization of ease of use consistent with ease of design and updating of expert information contained therein. These portions include presentation of a fixed, invariant inquiry sequence and storage matrices which are accessed on the basis of responses to the inquiry sequence, a portion in which a decision tree is traversed efficiently under control of an inference engine based on user selections, preferably from a menu or a hierarchy of menus and a portion in which the analysis sequence is carried out by parsing text from an expert database into queries which are also preferably in the form of menus. Capture of the decision trees traversed together with corresponding user input data for at least selected analyses performed by the Expert system provides for capture of additional rules and business practices and facilitates correction of determinations, consistency of user intervention and expansion of the expert system.

Abstract: A digital neural network architecture including a forward cascade of layers of neurons, having one input channel and one output channel, for forward processing of data examples that include many data packets. Backward cascade of layers of neurons, having one input channel and one output channel, for backward propagation learning of errors of the processed data examples. Each packet being of a given size. The forward cascade is adapted to be fed, through the input channel, with a succession of data examples and to deliver a succession of partially and fully processed data examples each consisting of a plurality of packets. The fully processed data examples are delivered through the one output channel. Each one of the layers is adapted to receive as input in its input channel a first number of data packets per time unit and to deliver as output in its output channel a second number of data packets per time unit.

Abstract: A signal processing system and method for accomplishing signal processing using a neural network that incorporates adaptive weight updating and adaptive pruning for tracking non-stationary signal is presented. The method updates the structural parameters of the neural network in principal component space (eigenspace) for every new available input sample. The non-stationary signal is recursively transformed into a matrix of eigenvectors with a corresponding matrix of eigenvalues. The method applies principal component pruning consisting of deleting the eigenmodes corresponding to the smallest saliencies, where a sum of the smallest saliencies is less than a predefined threshold level. Removing eigenmodes with low saliencies reduces the effective number of parameters and generally improves generalization. The output is then computed by using the remaining eigenmodes and the weights of the neural network are updated using adaptive filtering techniques.

Abstract: A system for generating an exception report for a particular problem based on call traffic. The system maintains a count of occurrences for that problem, happening within a sliding window of call traffic volume. When the count exceeds a threshold value, the system generates the exception report whereby appropriate action may be initiated to correct the problem.

Abstract: The accuracy of finding a cause from factors cited by means of a MYCIN method by narrowing down such factors is improved. With respect to deriving events which would not possibly occur if a particular factor 10C among a plurality of factors is a cause of abnormal event 100C, negative deriving events 12C and 1MC negating these deriving events of the other factors which would not occur with factor 10C being the cause are added to a fault tree chart as deriving events of factor 10C.

Abstract: A remote communication system for facilitating secure electronic sales of products, wherein a suitable seller input device in association with a local seller data transmission system couples the seller to a packet network system for communicating to a remote receiver/decoder apparatus to download registration data to the seller system. Upon retrieving a telephone toll number from the remote apparatus, a telcom network communication link for communicating the telephone number from the seller to the remote receiver allows the seller to upload the product that is to be offered for sale to the remote apparatus. The telcom connection, linking the seller system to the remote apparatus may also include a 900 number billing system whereby a fee is assessed and charged at a rate determined by the toll connection provider after uploading the product.

Abstract: A method of constructing and designing a fuzzy controller is developed based on converting a finite dimensional linear controller to a fuzzy controller. After the fuzzy controller is created, it can be enhanced using heuristic knowledge. The fuzzy controller is constructed or designed using a microprocessor to calculate a defuzzified output based on singleton fuzzification, product inferencing and centroidal defuzzification. A fuzzy controller can be created much more quickly and accurately than with previous methods.

Abstract: An ordering method for ranking membership function values (Wi) of linguistic input values (LWEi) in a fuzzy logic processor is presented The steps of the method are:a) within the time of one processor clock, successively reading the membership function values (Wi) into holding elements (L1 . . . L4) and, after every reading, outputs of the holding elements are through-connected onto outputs (A . . . D) of a selector (SC) as determined by a selection signal (SEL);b) comparing the signals at the outputs (A . . . D) of the selector in comparators (C1,C2,C3) and generating control signals (S1 . . . S3) for a unit (SELC) which in turn generates the new selection signal (SEL); andc) writing pointers (MAX, MAX', MAX", MIN', MIN) into position registers (P1 . . . P5) with the assistance of the control signals (S1 . . . S3) such that the pointers enable a ranked access to the membership function values (Wi) in the holding elements (L1 . . . L4).

Abstract: A fuzzy-logic-based evidence fusion tool for predicting function levels of a switch in a telecommunications network. Confidence factors related to the geographical location and to the physical characteristics of the switch are acquired and mapped into fuzzy membership values. The membership values activate a set of fuzzy evidence fusion rules at each function level, and a degree of truth for each rule is inferred. A defuzzification scheme combines the degree of truth of each rule into a single combined confidence factor for each function level. Finally, the maximum combined confidence factor for each rule is determined, and the switch function level corresponding to that confidence factor is assigned to the switch.

Abstract: An input value sorter for calculating a conformity grade in a fuzzy inferential device has a comparator for sequentially receiving data indicative of a position of each of vertexes of a membership function defined as a polygon and receiving fact data. The comparator sequentially compares the fact data showing a determined value with the data indicative of the position of each of the vertexes. The input value sorter also has a device for specifying a side or vertex of the polygon belonging to the fact data on the basis of compared results of the comparator and an input order of the vertexes. A grade arithmetic unit has this input value sorter. In this grade arithmetic unit, the membership function can be suitably sorted by a simplified construction.

Abstract: A rules-based method and apparatus for discriminating amongst a plurality of facts to arrive at a best result applies at least three primary decision-making rules, ie. (1) it is usually undesirable to use an input in combination with a fact which was based on that particular input; (2) using a fact based on several inputs is usually better than using a fact based on a single input; and (3) using a fact based on a shorter series of relationships is usually better than using a fact based on a longer series of relationships.

Abstract: A defuzzification device in which, depending on selection signals (FA1, FA2) and aggregated regulation weights (.gamma.), either in a maximum process a first/last maximum value at which an associated aggregated regulation weight is a maximum is selected as the sharp output value (g) and an activation signal (ENGARZ) is formed or, as an alternative, in a first or second center of gravity process (COG12) a sharp output value (g) is formed from center of gravity coordinates (S.sub..nu.) and aggregated regulation weights (.gamma..sub..nu.), alone or additionally using area measurement figures (F.sub..nu.), adders (ADD1, ADD2) with a register (Z-Reg, N-Reg) connected downstream in each case, a multiplier (MULT) and a divider (DIV) being able to be used advantageously both for the first and for the second center of gravity process, and one of the adders (ADD2) being used for the formation of the activation signal (ENGARZ), if a defuzzification is being carried out according to the maximum process.

Abstract: A fuzzy processor contains a fuzzifier which calculates from the digital input signal membership values of the input membership functions hit. The membership functions must be stored for this purpose. In order to reduce the memory requirements, in particular in conjunction with a high resolution of the input signals, it is expedient to describe the membership functions by means of shape-determining features such as restart points and slopes, and to store these features. In order to save calculating time, the definition range of the input signals is now divided into elementary intervals of equal size. During fuzzification, only that elementary interval is considered in which the input signal to be fuzzified is situated. This reduces the number of the membership functions to be considered. The result of the method is fuzzification of the input signals in conjunction with a minimum calculating time without the memory requirements for storing the membership functions becoming excessive.

Abstract: A unit in a computer for computing a closest distance between two three-dimensional convex objects. At least one of the objects is a convex curved surface object. A first closest vector arithmetic unit is provided for identifying a preliminary closest vector coupling the closest points of the two objects by repeatedly calculating for each object an inner product of a candidate for the preliminary closest vector directed from a first of the two objects to a second of the two objects and a plurality of position vectors. Each position vector being one of a position vector of a vertex where the first object is a convex polyhedron, and a position vector of a control point of a convex curved surface patch where the first object is a convex curved surface object comprising sets of convex curved surface patches. A position vector for each object having a maximum inner product is extracted and a new candidate for the preliminary closest vector is formed by a vector connecting the two position vectors.

Abstract: A radio frequency manager (12) for choosing a selective call transmitter (20 or 25) among a plurality of selective call transmitters comprises a memory (13) for storing historical data on messages previously sent, the historical data comprising for each transmitter at least a latency measurement and a message queue profile and another memory (15) for storing pending messages, each of the pending messages having current information comprising at least an associated priority for the pending messages. The radio frequency manager further comprises a sorter (21) for classifying the historical data and the current information using fuzzy logic, wherein the sorter comprises a fuzzifier (14) for generating membership values, a rule base (16) for applying a set of predetermined rules to the current input conditions to perform a mapping with the historical conditions, and a defuzzifier (18) for generating a non-fuzzy prediction recommendation from the mapping for recommending the optimum transmitter.

Abstract: A robot includes a robot body to which the procedure of a task is teachable, a teaching portion, an operating portion, a control portion, and a changeover portion. The teaching portion includes a signal generating portion for generating a drive signal for driving the robot body, and a deadman switch which indicates by its opened/closed state whether the procedure of the task is being taught to the robot body or not. The operating portion generates a start signal for starting the robot body. The control portion receives the drive signal and a signal indicating the opened/closed state of the deadman from the teaching portion and the start signal from the operating portion. The changeover portion operates so that the start signal from the operating portion is prevented from being supplied to the control portion when the teaching portion is electrically connected to the control portion.

Abstract: A method and system are disclosed for efficiently correlating a plurality of events within a data processing system to determine an action to be performed. According to the present invention, a rule network is provided which includes a number of objects arranged in a tree structure having at least one parent object and a plurality of child objects, where a parent object is logically linked to a child object by passing an input event received by the parent object to the child object. At least one of the objects within the rule network invokes an action in response to receipt of a particular input event. In response to receipt of a series of events, the series of events is evaluated through the tree structure rule network. In response to the evaluation, an action is invoked at one of the number of objects.