Quantitative Revision of Scientific Models
Lecture notes in computer sciencePublished 23 August 2007
Kazumi Saito, Pat Langley
Citations1
SJR quartileQ2
SJR score0.35
SNIP0.55
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
This chapter describes an approach to improving scientific models that are cast as sets of equations, then recounts its application to revising parameter values, intrinsic properties, and functional forms, in each case achieving reduction in error on Earth science data while retaining the communicability of the original model.
Abstract
Research on the computational discovery of numeric equations has focused on constructing laws from scratch, whereas work on theory revision has emphasized qualitative knowledge. In this chapter, we describe an approach to improving scientific models that are cast as sets of equations. We review one such model for aspects of the Earth ecosystem, then recount its application to revising parameter values, intrinsic properties, and functional forms, in each case achieving reduction in error on Earth science data while retaining the communicability of the original model. After this, we consider earlier work on computational scientific discovery and theory revision, then close with suggestions for future research on this topic.
Keywords
Computer ScienceEnvironmental Science
Medical Entomology and ZoologyScientific Discovery: Computational Explorations of the Creative Processes
812 Citations1987Pat Langley, Herbert A. Simon +2 more
Neural ComputationProduct Units: A Computationally Powerful and Biologically Plausible Extension to Backpropagation Networks
417 Citations1989Richard Durbin, David E. Rumelhart
A new form of computational unit for feedforward learning networks of the backpropagation type that calculates a weighted product, where each input is raised to a power determined by a variable weight.
National Conference on Artificial IntelligenceChanging the rules: a comprehensive approach to theory refinement
129 Citations1990Dirk Ourston, Raymond J. Mooney
The approach uses explanations to focus the corrections to the theory, with the corrections being supplied by an inductive component to preserve the structure of the original theory as much as possible.
Contemporary mathematics - American Mathematical SocietyAutomated Theory Formation in Mathematics
118 Citations1984Douglas B. Lenat
A program called "AM" is described which rediscovers hundreds of common concepts, including singleton sets, natural numbers, arithmetic and theorems and provides a definite but immense space which AM begins to explore.
Declarative Bias in Equation Discovery
112 Citations1997Ljupčo Todorovski, Sašo Džeroski
Minds at UW (University of Wisconsin)Symbolic knowledge and neural networks: insertion, refinement and extraction
99 Citations1992Geoffrey G. Towell
A hybrid system called K scBANN (Knowledge-Based Artificial Neural Networks) is a three-part hybrid learning system built on top of "neural" learning techniques which is shown to be an effective combination of these two learning methods.
Tellus BGlobal model estimates of carbon and nitrogen storage in litter and soil pools: response to changes in vegetation quality and biomass allocation
98 Citations1997Christopher Potter, Steven Klooster
Machine LearningDetermining Arguments of Invariant Functional Descriptions
72 Citations1986Mieczyslaw M. Kokar
This paper presents a method, called COPER, for discovering invariant functional descriptions, which eliminates irrelevant arguments, generates additional relevant arguments, and generates a functional formula.
Automated discovery in a chemistry laboratory
65 Citations1990Jan M. Żytkow, Jieming Zhu +1 more
After minor adjustments FAHRENHEIT has been able to discover regularities in maxima locations and heights, and to analyse repeatability of measurements by the same mechanism, adapted from BACON, by which all numerical regularities are detected.
Global Biogeochemical CyclesInterannual variability in soil trace gas(CO<sub>2</sub>, N<sub>2</sub>O, NO) fluxes and analysis of controllers on regional to global scales
56 Citations1998Christopher Potter, Steven Klooster
Lecture notes in computer scienceThe Computer-Aided Discovery of Scientific Knowledge
55 Citations1998Pat Langley
This paper describes five historical stages of the scientific discovery process, which it uses as an organizational framework in describing applications, and identifies five distinct steps during which developers or users can influence the behavior of a computational discovery system.
International Joint Conference on Artificial IntelligenceRediscovering physics with BACON.3
54 Citations1979Pat Langley
BACON.3 has shown its generality by rediscovering versions of the Ideal gas law, Kepler's third law, Coulomb's law, Ohm'slaw, and Galileo's laws for the pendulum and constant acceleration.
Law discovery using neural networks
52 Citations1997Kazumi Saito, Ryohei Nakano
The main advantage of this 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.
Lecture notes in computer scienceTheory Revision in Equation Discovery
20 Citations2001Ljupčo Todorovski, Sašo Džeroski
This paper proposes a theory revision method suitable for use with the equation discovery system Lagramge and shows that small changes in the model parameters and structure considerably improve the accuracy of the model.
Discovering admissible simultaneous equations of large scale systems
15 Citations1998Takashi Washio, Hiroshi Motoda
The theoretical foundations of the structure identification and the algorithm of SSF are described, and its efficiency and practicality are demonstrated and discussed with large scale working examples.
Discovering Communicable Scientific Knowledge from Spatio-Temporal Data
15 Citations2001Mark Schwabacher, Pat Langley
This paper describes how regression rules were used to improve upon a result previously published in the Earth science literature and describes two visualization techniques that were developed to aid in understanding the model by exploiting the spatial nature of the data.
Lecture notes in computer scienceDiscovery of Nominally Conditioned Polynomials using Neural Networks, Vector Quantizers and Decision Trees
11 Citations2000Kazumi Saito, Ryohei Nakano
This paper proposes a new version of RF6 by greatly improving its procedure of restoring nominally conditioned polynomials from a trained neural network, which will suffer from a combinatorial explosion in the process of restoring rules from atrained neural network.
A Divide and Conquer Approach to Learning from Prior Knowledge
7 Citations2000Eric Chown, Thomas G. Dietterich
A new divide-and-conquer method is presented that analyzes the model to identify a series of smaller optimization problems whose sequential solution solves the global calibration problem.