Keywords: web application, equation discovery, scientific gateway, grid, Lagramge, peak ground acceleration.

Equation discovery (ED) is an emerging area of machine learning, where the main objective is to find a proper equation structure that best describes the dependencies between variables in a given data set. In ED, the key add-on is the method which is used to generate equation structures as hypothetical models of the investigated phenomenon. The Lagramge system uses a context-free grammar (CFG) to generate hypothetical equation structures which are presented as input information to the system [1]. The CFG is a formalism, which makes it possible to provide domain knowledge to the ED system and so guide it towards more plausible equation formulations.

Recent studies have shown that the Lagramge ED system can be useful in the engineering disciplines in general and earthquake engineering in particular [2]. Since the implemented algorithm is computationally intensive in nature it is beneficial to use a grid infrastructure. The academic and research network of Slovenia provides and maintains the national grid infrastructure under the Slovene national grid initiative (SLING). The researchers can use a total of 4,268 processors and associated storage facilities for various high-performance computing experiments.

The goal of the present study was to develop a web application and associated scripts that would make it possible to execute the experiments with the Lagramge ED system on the grid, in our case the SLING's infrastructure. The whole process consists of three separate phases:

1. the preparation of the job,
2. the overview of the running jobs; and
3. the acquisition of the results.

The procedure of the job preparation is based on the required information for the Lagramge ED system, provided in [1] and the readme file enclosed with Lagramge source files.

An example from the domain of earthquake engineering is presented as in [2]. In this example, the equation models for the earthquake's peak ground acceleration are induced, which are required to design seismically safe structures. The web application presented for equation discovery can also be used in other engineering disciplines, since equations are commonly used as models of natural phenomena.

1

L. Todorovski, S. Dzeroski, "Declarative bias in equation discovery", in "Proceedings of the 14th International Conference on Machine Learning", Morgan Kaufmann, 376-384, 1997.

2

Š. Markic, V. Stankovski, "An Equation-Discovery Approach to Earthquake-Ground-Motion Prediction", Engineering Applications of Artificial Intelligence, in press, doi:10.1016/j.engappai.2012.12.005

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £40 +P&P)