GB2258554A - Function-neuron net analyzer - Google Patents

Function-neuron net analyzer Download PDF

Info

Publication number
GB2258554A
GB2258554A GB9117059A GB9117059A GB2258554A GB 2258554 A GB2258554 A GB 2258554A GB 9117059 A GB9117059 A GB 9117059A GB 9117059 A GB9117059 A GB 9117059A GB 2258554 A GB2258554 A GB 2258554A
Authority
GB
United Kingdom
Prior art keywords
function
neuron
neuron net
net
control
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
GB9117059A
Other versions
GB9117059D0 (en
Inventor
Haneef Akhter Fatmi
Ching-Cheng Lee
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Priority to GB9117059A priority Critical patent/GB2258554A/en
Publication of GB9117059D0 publication Critical patent/GB9117059D0/en
Publication of GB2258554A publication Critical patent/GB2258554A/en
Withdrawn legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
    • G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means

Abstract

A function-neuron chip has a processing analyzer function-neuron 3 with massive parallelism for real-time processing, a feedback controller function-neuron 5 to control the feedback information for correction and improvement of output, and an adaptive analyzer function neuron 2 for the net to learn so that it can be applied to complex problems in engineering and physics computations, real-time applications in command and control as well as perception recognition including speech, language, and visual images. <IMAGE>

Description

FUNCTION-NEURON NET ANALYZER This invention relates to a function-neuron net analyzer.
Function-neuron net analyzer is a chip of neural net which comprises a set of special purpose function-neurons and is capable of analyzing complex applications which require high speed processing and learning, such as engineering computation, computational physics, supercomputer algorithms, and artificial perception including speech, language, and images.
The current analyzers on the market are either conventional software/hardware package with no learning capability or made of simple neural-type devices with limited parallelism with no real-time capability.
According to the present invention, there is provided a net of function neurons, each of which is made of either a set of function-neurons or atomic elements (neurons) with shared-RAM to perform a specialized function.
There are three main function-neurons in the net: an adaptive analyzer, a processing analyzer with massive parallelism, and a feedback controller to provide learning for output correction and improvement.
Figure 1 shows in perspective, a function-neuron net.
Figure 2 illustrates in perspective, the element structure of a function-neuron.
Figure 3 illustrate in perspective, another super-structure of a function-neuron consisting of the function-neurons shown in Fig 2 as its elements.
Referring to Fig 1, a function-neuron net chip has sensor or problem input to be analyzed labelled 1, adaptive analyzer function-neuron labelled 2, processing analyzer function-neuron labelled 3, output labelled 4, feedback controller function-neuron labelled 5, and adaptive analyzer indicator labelled 11.
There are two types of function neurons. One is the elementary functionneuron shown in Fig 2 and labelled 6 has RAM labeled 7, and element neuron labeled 8. The adaptive analyzer and feedback controller belong to this type of function-neuron.
The other type of function-neuron is shown in Fig 3 and labeled 9 has the function neuron labelled 10 (same structure as Fig 2) as its elements.
Processing analyzer belongs to this type of function-neuron so that it can provide massive parallelism to analyze problem in real-time.
In order to analyze the problem or the artificial perception, the input will be fed to this chip and a indicator labelled 11 outside the adaptive analyzer will signal the completion after several times of feedback control loop to remove some of the ambuguities. The output labelled 4 will then have the final answer or perception output.

Claims (5)

ClAIMS
1. A chip of function-neuron net has n-dimensional input and output, and can be used as an analyzer for linear, nonlinear, and hierarchical inputs including engineering and physics computations, supercomputer algorithm analysis and architecture porting, artificial intelligence applications, and perception recognition such as speech, language, vision, etc.
2. A function-neuron net can be used as a real-time device for communication and command control system.
3. A function-neuron net can be used to produce VLSI chips for constructing a more general powerful neural computer to process tasks of numerical, nonnumerical, and perception problems.
4. A function-neuron net can be used for developing scheduling control strategies in computer systems as well as process control of manufacturing plants.
5. Function-neuron net chips will be able to make optical computers.
6 Function-neuron net chips will be able to make a control device to monitor networks of computers domestically and internationally for control and safety.
GB9117059A 1991-08-07 1991-08-07 Function-neuron net analyzer Withdrawn GB2258554A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
GB9117059A GB2258554A (en) 1991-08-07 1991-08-07 Function-neuron net analyzer

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
GB9117059A GB2258554A (en) 1991-08-07 1991-08-07 Function-neuron net analyzer

Publications (2)

Publication Number Publication Date
GB9117059D0 GB9117059D0 (en) 1991-09-18
GB2258554A true GB2258554A (en) 1993-02-10

Family

ID=10699649

Family Applications (1)

Application Number Title Priority Date Filing Date
GB9117059A Withdrawn GB2258554A (en) 1991-08-07 1991-08-07 Function-neuron net analyzer

Country Status (1)

Country Link
GB (1) GB2258554A (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1988007234A1 (en) * 1987-03-12 1988-09-22 Analog Intelligence Corporation Back propagation system
WO1990010274A1 (en) * 1989-02-21 1990-09-07 Rijksuniversiteit Te Leiden Neuronal data processing network
GB2239972A (en) * 1990-01-11 1991-07-17 British Telecomm Neural network

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1988007234A1 (en) * 1987-03-12 1988-09-22 Analog Intelligence Corporation Back propagation system
WO1990010274A1 (en) * 1989-02-21 1990-09-07 Rijksuniversiteit Te Leiden Neuronal data processing network
GB2239972A (en) * 1990-01-11 1991-07-17 British Telecomm Neural network

Also Published As

Publication number Publication date
GB9117059D0 (en) 1991-09-18

Similar Documents

Publication Publication Date Title
Sun et al. Green's function method for fast on-line learning algorithm of recurrent neural networks
Zomaya Reinforcement learning for the adaptive control of nonlinear systems
Park et al. A neuro-genetic controller for nonminimum phase systems
GB2258554A (en) Function-neuron net analyzer
Keibek et al. A fast partitioning algorithm and a comparison of binary feedforward neural networks
Zhang et al. Key technologies of smart factory machine vision based on efficient deep network model
Zhu et al. Lagrange neural networks for linear programming
Villalobos et al. Learning capability assessment and feature space optimization for higher-order neural networks
Lebart Correspondence analysis, discrimination, and neural networks
Zhang et al. Study of nonlinear system identification based on support vector machine
Li et al. Non-negative matrix-set factorization
Peper et al. A symmetric linear neural network that learns principal components and their variances
Cancelo et al. A parallel analog signal processing unit based on radial basis function networks
Patel et al. A Review on Facial Emotion Recognition That Uses Machine Learning Algorithms
US20240054335A1 (en) Ai-based pattern identification transformer
Chen et al. Neural network automata
Prasad et al. Intelligent robot used in the field of practical application of artificial neural network & machine vision
Salazar-Colores et al. A fast and robust deep learning approach for hand object grasping confirmation
Courrieu Solving time of least square systems in Sigma-Pi unit networks
Pandey et al. Handwritten Text Conversion By Using ANN Approach
Stöckel et al. A Geometric Interpretation of Feedback Alignment.
Elmoznino et al. Visual representations derived from multiplicative interactions
TR201806540A2 (en) Critical Network Training Method for Deep Artificial Neural Networks in Machine Learning
WO2023117473A1 (en) Apparatus and method for object detection/recognition
JPH08249303A (en) Signal processor

Legal Events

Date Code Title Description
WAP Application withdrawn, taken to be withdrawn or refused ** after publication under section 16(1)