Law discovery using neural networks

Authors:
Kazumi Saito;Ryohei Nakano
Affiliations:
NTT Communication Science Laboratories, Soraku-gun, Kyoto, Japan;NTT Communication Science Laboratories, Soraku-gun, Kyoto, Japan
Venue:
IJCAI'97 Proceedings of the Fifteenth international joint conference on Artifical intelligence - Volume 2
Year:
1997

Citing 11
Cited 11

Scientific discovery: computational explorations of the creative process

Scientific discovery: computational explorations of the creative process
Data-driven approaches to empirical discovery

Artificial Intelligence
Product units: a computationally powerful and biologically plausible extension to backpropagation networks

Neural Computation
A robust approach to numeric discovery

Proceedings of the seventh international conference (1990) on Machine learning
Learning internal representations by error propagation

Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
Basis-function trees as a generalization of local variable selection methods for function approximation

NIPS-3 Proceedings of the 1990 conference on Advances in neural information processing systems 3
Integrating quantitative and qualitative discovery in the ABACUS system

Machine learning
First- and second-order methods for learning: between steepest descent and Newton's method

Neural Computation
Partial BFGS update and efficient step-length calculation for three-layer neural networks

Neural Computation
Stochastic Complexity in Statistical Inquiry Theory

Stochastic Complexity in Statistical Inquiry Theory
Bivariate Scientific Function Finding in a Sampled, Real-Data Testbed

Machine Learning

Extracting regression rules from neural networks

Neural Networks
Discovery of Relevant Weights by Minimizing Cross-Validation Error

PADKK '00 Proceedings of the 4th Pacific-Asia Conference on Knowledge Discovery and Data Mining, Current Issues and New Applications
Computational Characteristics of Law Discovery Using Neural Networks

DS '98 Proceedings of the First International Conference on Discovery Science
Discovery of a Set of Nominally Conditioned Polynomials

DS '99 Proceedings of the Second International Conference on Discovery Science
Discovering Polynomials to Fit Multivariate Data Having Numeric and Nominal Variables

Progress in Discovery Science, Final Report of the Japanese Discovery Science Project
Computational Revision of Quantitative Scientific Models

DS '01 Proceedings of the 4th International Conference on Discovery Science
Finding Polynomials to Fit Multivariate Data Having Numeric and Nominal Variables

IDA '01 Proceedings of the 4th International Conference on Advances in Intelligent Data Analysis
Quantitative Revision of Scientific Models

Computational Discovery of Scientific Knowledge
Convergence of BP algorithm for product unit neural networks with exponential weights

Neurocomputing
Nominally piecewise multiple regression using a four-layer perceptron

KES'07/WIRN'07 Proceedings of the 11th international conference, KES 2007 and XVII Italian workshop on neural networks conference on Knowledge-based intelligent information and engineering systems: Part III
An information-based approach towards neuro-evolution

Proceedings of the 14th annual conference companion on Genetic and evolutionary computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes a new connectionist approach to numeric law discovery; i.e., neural networks (law-candidates) are trained by using a newly invented second-order learning algor ithm based on a quasi-Newton method, called BPQ, and the Minimum Description Length criterion selects the most suitable from lawcandidates. The main advantage of our method over previous work of symbolic or connectionist approach is that it can efficiently discover numeric laws whose power values are not restricted to integers. Experiments showed that the proposed method works well in discovering such laws even from data containing irrelevant variables or a small amount of noise.