‘Neural Net Simulation on Transputers
نویسندگان
چکیده
2 Neural Nets, Statistical Physics and Optimization It has recently been shown that neural net models can be based upon the principles of statistical physics, In particular the spin sociative memories and optimization networks. The simulation of such neural networks promises to provide valuable insights into human and machine vision. However, for a large amay, the calculations involved become prohibitive on conventional serial machines. glass models of magnetic lattices have been used to design asThe relationship between neural net models and the physics of large scale magnetic spin systems has become widely appreciated since the work of Hopfield illThe magnetic interaction between
منابع مشابه
A Flexible Distributed Simulator for Object-Oriented Petri Nets
Petri nets have been used to model and analyse distributed systems. Highlevel nets such as coloured nets and object-oriented nets increase the descriptive power by allowing the tokens which flow through the net to be associated with arbitrary data values. Algorithms have been specified for the distributed simulation of simple petri nets on a set of transputers. This paper considers the extensio...
متن کاملParallel Implementations of Neural Network Simulations
At the present time, backpropagation is the most popular learning algorithm for multilayer feedforward neural networks. It can be expressed in the formalism of vector or matrix algebra : extensions of scalar operations and matrix products. These operations can be performed with a high degree of parallelism. Three architectures involving loosely coupled processors are considered: torus, mesh and...
متن کاملThe Identification of Hippocampal Network Function
This paper presents an approach to the identification of models characterizing the functional properties of the hippocampal formation. In this approach the network properties are characterized as the composite of input/output functions measured for each subsystem in the network. The characterizing functions are the kernels of a functional power series. The calculation and interpretation of thes...
متن کاملA Survey and Experimental Study of Neural Network AUV Control
This paper presents a survey of neural network controllers for AUVs. A direct adaptive neural net controller incorporating integral action was designed for the heave motion of the ODIN underwater vehicle. The neural net controller is trained on-line by parallel recursive error prediction method and the critic equation. The influence of the various design parameters of the neural net controller ...
متن کاملRandomized PRAM Simulation Using T Transputers
The parallel random access machine PRAM is the most commonly used general purpose machine model for describing parallel computations Unfortunately the PRAM model is not phy sically realizable since on large machines a parallel shared memory access can only be accompli shed at the cost of a signi cant time delay A number of PRAM simulation algorithms have been presented in the literature The alg...
متن کامل