The pi-sigma network: an efficient higher-order neural network for pattern classification and function approximation

Published 9 December 2002

Yoan Shin, Joydeep Ghosh

DOI10.1109/ijcnn.1991.155142

Citations302

TL;DR

Simulation results show good convergence properties and accuracy for function approximation, and Comparative results using the DARPA acoustic transient data set are provided to highlight the classification abilities of pi-sigma networks.

Abstract

Introduces a novel feedforward network called the pi-sigma network. This network utilizes product cells as the output units to indirectly incorporate the capabilities of higher-order networks while using a fewer number of weights and processing units. The network has a regular structure, exhibits much faster learning, and is amenable to the incremental addition of units to attain a desired level of complexity. Simulation results show good convergence properties and accuracy for function approximation. Comparative results using the DARPA acoustic transient data set are also provided to highlight the classification abilities of pi-sigma networks.>

Keywords

Computer Science

IEEE ASSP MagazineAn introduction to computing with neural nets

7,858 Citations1987Richard P. Lippmann

The MIT Press eBooksParallel Distributed Processing

3,227 Citations1987James L. McClelland, David E. Rumelhart +1 more

Proceedings of the IEEE30 years of adaptive neural networks: perceptron, Madaline, and backpropagation

2,242 Citations1990Bernard Widrow, Michael A. Lehr

The history, origination, operating characteristics, and basic theory of several supervised neural-network training algorithms (including the perceptron rule, the least-mean-square algorithm, three Madaline rules, and the backpropagation technique) are described.

IEEE Communications MagazinePattern classification using neural networks

812 Citations1989Richard P. Lippmann

The author extends a previous review and focuses on feed-forward neural-net classifiers for static patterns with continuous-valued inputs, examining probabilistic, hyperplane, kernel, and exemplar classifiers.

Applied OpticsLearning, invariance, and generalization in high-order neural networks

676 Citations1987C. Lee Giles, Tom Maxwell

High-order neural networks have been shown to have impressive computational, storage, and learning capabilities because the order or structure of a high- order neural network can be tailored to the order of a problem.

Biological CyberneticsOn optimal nonlinear associative recall

139 Citations1975Tomaso Poggio

The problem of determining the nonlinear function (“blackbox”) which optimally associates two sets of data is considered and an iteration method based on the concept of the generalized inverse of a matrix provides the polynomial mapping of degreek onX by whichY is retrieved in an optimal way in the least squares sense.

Rapid training of higher-order neural networks for invariant pattern recognition

71 Citations1989Rei, Spirkovska +1 more

Initial results show higher order neural networks to be vastly superior to multilevel first-order networks trained by backpropagation for applications where invariant pattern recognition is required.

Connectivity strategies for higher-order neural networks applied to pattern recognition

35 Citations1990Lilly Spirkovska, Max B. Reid

Different strategies for non-fully connected HONNs (higher-order neural networks) are discussed, showing that by using such strategies an input field of 128×128 pixels can be attained while still achieving in-plane rotation and translation-invariant recognition.