Please use this identifier to cite or link to this item: http://hdl.handle.net/10216/67130
Author(s): Francisco Reinaldo
Marcus Siqueira
Rui Camacho
Luís Paulo Reis
Title: A tool for Multi-Strategy Learning
Issue Date: 2006
Abstract: This paper presents the AFRANCI tool for the developmentof Multi-Strategy learning systems. AFRANCI allows users to build, inan interactive and easy way, complex systems. Systems are built using atwo step methodology: design of the structure of the system; and fill inthe modules. The structure of the target system is a collection of interconnectedmodules. The user may then choose among a variety of learningalgorithms to construct each module. The tool has several built-in MachineLearning algorithms and interfaces that enable it to use externallearning tools like WEKA or CN2. AFRANCI uses the interdependencyof the modules to determine the sequence of their training. To improveusability, the tool uses a wrapper that hides from the user the parametertuning procedure for each algorithm. In a final step of the designsequence AFRANCI generates a compact and legible ready-to-use ANSIC++ open-source code for the final system.To illustrate the concept we have empirically evaluated the tool in thecontext of the RoboCup Rescue domain. We have developed a smallsystem that uses both neural networks and rules in the same system.The experiment have shown that a very significant speed up is attainedin the development of systems when using this tool.
Subject: Tecnologia de computadores, Engenharia electrotécnica, electrónica e informática
Call Number: 69530
URI: http://hdl.handle.net/10216/67130
Document Type: Artigo em Revista Científica Internacional
Rights: openAccess
License: https://creativecommons.org/licenses/by-nc/4.0/
Appears in Collections:FEUP - Artigo em Revista Científica Internacional

Files in This Item:
File Description SizeFormat 
69530.pdfA tool for Multi-Strategy Learning373.35 kBAdobe PDFThumbnail
View/Open


This item is licensed under a Creative Commons License Creative Commons