Data processing: artificial intelligence – Machine learning
Reexamination Certificate
2001-07-05
2004-02-24
Davis, George B. (Department: 2121)
Data processing: artificial intelligence
Machine learning
C706S011000, C706S014000
Reexamination Certificate
active
06697789
ABSTRACT:
BACKGROUND OF THE INVENTION
The invention concerns a computer system used to determine and evaluate data or groups of data, each of which includes a specific characteristic, particularly for simulation of human perception via sense organs.
In robotics, one is concerned with computer systems that deal with handling processes based on external or internal data with the help of a specific program, and that pass the calculation results on to actuators, effectors, or generally to a motor control unit, for example, so that work is performed by the robot automatically and independently. A further goal is to build “learning” robots, i.e., to configure the computer system so that that it expands or restructures existing programming based on external data in order to optimize the assigned task and more effectively perform it.
Examples for such computer systems are neural net computers that simulate the processes in regions of the human brain based on external stimulation, and that produce corresponding calculation results.
German patent DE-C2-3707998, of which the applicant is a co-inventor, makes a step in this direction. In the computer system presented there, an attempt is made to at least partially emulate the neuron brain structure. For this, the computer system includes a number of node computers that correspond to neurons in the human brain, and bi-directional information conductors connecting the node computers that correspond to dendrites in the human brain. This computer serves to calculate the environment-related self-awareness, and is organized permutographically, whereby, in principle, each node computer can assume the command control of all computation operations of the entire computer. This achieves a degree of potential command execution redundancy that has actually been observed for neurons in the human brain.
Along with this permutographically-organized computer, an additional keno-grammatically-organized computer is provided that communicates with the permutogaph-computer via compilers. The kenograph computer simulates the neuroglia that have been proved to be responsible for information transfer between individual neurons and for the observed restructuring in the neuron aggregate.
The calculation results achieved by this complex computer system are represented as a path through the computer system.
A particular problem in “learning” robots is the simulation of perceptions via sense organs. If a computer system with a corresponding structure and program that simulated human perception were available here, the evaluation of the output signals from sensors, whether they are optical, touch, or scent sensors etc., might be greatly simplified.
SUMMARY OF THE INVENTION
The present invention is based on the objective of creating a computer system of the type mentioned above that evaluates data or groups of data possessing particular characteristics so that possible responses, e.g., handling instructions for the motor systems of a robot, may be selected and evaluated in a meaningful manner.
This objective is achieved by the invention by means of a computer system with a receptor field, a processor field, a switching field, a phase circuit, and a phase program circuit, each of which is specifically configured.
The receptor field includes sensors, each of which corresponds to data with a specific characteristic (e.g., characteristics or properties A, B, or C), and which then issues an output signal characterizing this specific property. This corresponds to the “key/keyhole principle” often observed in nature, whereby a receptor, in this case the sensor, features a material/chemical structure as the “keyhole” into which only a complementary structure, the data in the simulation with the specific characteristic, will fit as the key.
The sensors are connected with links, e.g., wires, with a processor field consisting of processors; in the simulation the processors correspond to neurons in the human brain, and the connections between sensors and processors correspond to the afferent axons that lead from each sense organ to the central nervous system. This connection is a 1:1 connection i.e., each sensor is assigned to a processor or processor group.
The switching field switches the existing connections between sensors and processor on or off. This switching field corresponds in the simulation to the oligodendrozytes that surround the myelinscheiden with their arms, and can limit or release them.
By means of the phase circuit, the processors or a processor group and the switching field may be activated and deactivated in cycled time phases. In the simulation, this phase circuit corresponds to the astrozytes with their arms. The connection circuit and the phase circuit usually operate synchronously, so that a redundant release or blockage of the lines results. It is also possible that each phase circuit specifies a program section that is the basis for a common intention of the evaluations to be performed. The pertinent time phase is relatively long. With the specification, the corresponding processors responsible for the evaluation are released. The connection circuit has the task of releasing the processors responsible for the individual sections within the specified program sections in several sections, i.e., shorter time phases.
The phase program circuit controls the computer system and determines the progression of the time phases and thereby the limitation or release of the processor and the switching field associated with it by means of the phase control. The phase program circuit and some functions of the phase circuit correspond in the simulation to the glia cells that have access to an experimentally-ensured internally-generated pulsation that corresponds to the time phases queried. During a time phase, a certain combination of queried sensors, i.e., in the simulation, a combination of statuses of the receptors, is sought and checked. This “search” corresponds to the known intentional, behavior-related perception. Switching of the time phases of the phase program circuit and the connections occurs in parallel, a reference to the hypothesis that the glia possesses a spatial and temporal limit-setting function.
The temporally-cycled phase program circuit specifies the data with specific characteristics that it anticipates in the receptor field for a certain intention, for example, certain colors, shapes, etc. The data actually present at the sensors are compared with the specified data and analyzed, for example, to determine whether the anticipated data are present as distributed at the sensors, and at which sensors they are actually present.
The phase circuit activates just those processors that are provided for information processing of the anticipated phase program.
The switching field activates just those sensors that are dictated by the phase program.
Each pertinent phase program applies both to the phase circuit and to the switching field, so that these two devices are connected in parallel.
By means of the phase program circuit, for example, a primary, secondary, or tertiary analysis of the status of the receptors may result with the support of the kenogrammatics.
A primary analysis gives a response to the question whether sensors have been queried at all by the properties specified by the phase program, i.e., by specific data. For a robot, this represents a response to the query whether it is even located within a suitable environment within which a chance exists that the intended phase program may be implemented with the proper handling instructions.
A secondary analysis gives a response to the questions of how many sensors are being queried by the specific characteristics intended by the phase program, and how these sensors are distributed within the receptor field.
A tertiary analysis gives a response to the question where the sensors are positioned within the receptor field that are being queried by the specific data intended by the phase program.
Based on this, and possibly other, analyses, the computer system can determine a distribution of specific data. This
LandOfFree
Computer system, particularly for simulation of human... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Computer system, particularly for simulation of human..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Computer system, particularly for simulation of human... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3339733