Extracting propositions from trained neural networks

Authors:
Hiroshi Tsukimoto
Affiliations:
Research & Development Center, Toshiba Corporation, Kawasaki, Japan
Venue:
IJCAI'97 Proceedings of the Fifteenth international joint conference on Artifical intelligence - Volume 2
Year:
1997

Citing 6
Cited 2

C4.5: programs for machine learning

C4.5: programs for machine learning
Constant depth circuits, Fourier transform, and learnability

Journal of the ACM (JACM)
Combining Symbolic and Neural Learning

Machine Learning
The discovery of propositions in noisy data

Machine intelligence 13
Structural learning with forgetting

Neural Networks
Efficient algorithms for inductive learning—an application of multi-linear functions to inductive learning

Machine intelligence 14

Rule Extraction from Prediction Models

PAKDD '99 Proceedings of the Third Pacific-Asia Conference on Methodologies for Knowledge Discovery and Data Mining
The functional localization of neural networks using genetic algorithms

Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents an algorithm for extracting propositions from trained neural networks. The algorithm is a decompositional approach which can be applied to any neural network whose output function is monotone such as sigmoid function. Therefore, the algorithm can be applied to multi-layer neural networks, recurrent neural networks and so on. The algorithm does not depend on training methods. The algorithm is polynomial in computational complexity. The basic idea is that the units of neural networks are approximated by Boolean functions. But the computational complexity of the approximation is exponential, so a polynomial algorithm is presented. The authors have applied the algorithm to several problems to extract understandable and accurate propositions. This paper shows the results for votes data and mushroom data. The algorithm is extended to the continuous domain, where extracted propositions are continuous Boolean functions. Roughly speaking, the representation by continuous Boolean functions means the representation using conjunction, disjunction, direct proportion and reverse proportion. This paper shows the results for iris data.