Logo
Distributed Genetic Programming Framework
Home | Developers | Downloads | Source | Team
print print

Documents

This page contains the documents published related to the Distributed Genetic Programming Framework project.
ID Date Author(s) Title/Link Content
014 2006-10-13 Thomas Weise DGPF Flyer A german language flyer for the DGPF project, pinned onto our blackbords outside of the office in order to lure students into the project.
013 2006-10-10 Thomas Weise, Kurt Geihs BIOMA 2006 Paper The paper presented in BIOMA 2006 - The 2nd International Conference on Bioinspired Optimization Methods and their Application in Ljubljana (Slovenia) (using presentation 012) at the Jožef Stefan International Postgraduate School on October the 10th, 2006.. BIOMA is a small but really good conference, properly organized giving people the opportunity to talk with each other and to listen to all presentations in a good atmosphere. Respect.
@conference{WG2006DGPFc,
  author        = {Thomas Weise, Kurt Geihs},
  title         = {DGPF - An Adaptable Framework for Distributed Multi-Objective
                   Search Algorithms Applied to the Genetic Programming of Sensor
                   Networks},
  month         = {Oct},
  year          = {2006},
  abstract      = {We present DGPF, a framework providing multi-objective, auto-adaptive
                   search algorithms with a focus on Genetic Programming. We first introduce
                   a Common Search API, suitable to explore arbitrary problem spaces with
                   different search algorithms. Using our implementation of Genetic
                   Algorithms as an example, we elaborate on the distribution utilities
                   of the framework which enable local, Master/Slave, Peer-To-Peer, and
                   P2P/MS hybrid distributed search execution. We also discuss how
                   heterogeneous searches consisting of multiple, cooperative search
                   algorithms can be constructed. Sensor networks are distributed systems
                   of nodes with scarce resources. We demonstrate how Genetic Programming
                   based on our framework can be applied to create algorithms for sensor
                   nodes that use these resources very efficiently.},
  keywords      = {Genetic Programming, Genetic Algorithms, DGPF, Sensor Network, Sensor Node},
  URL           = {http://dgpf.sourceforge.net/documents/013-2006-10-10-bioma_paper},
  pages         = {157--166},
  type          = {Resaerch Talk Paper},
  affiliation   = {University of Kassel,
                  Distributed Systems Group},
  location      = {Jo\v{z}ef Stefan Institute, Ljubljana, Slovenia},
  language      = {en},
  rights        = {unrestricted},
  booktitle     = {Proceedings of the Second International Conference on Bioinspired Optimization
                   Methods and their Application, BIOMA 2006},
  ISBN          = {ISBN-10: 961-6303-81-3, ISBN 13: 978-961-6303-81-1},
  editor        = {Bogdan Filipi\v{c} and Jurij \v{S}ilc},
  publisher     = {Jo\v{z}ef Stefan Institute, Ljubljana, Slovenia},
  series        = {International Conference on Bioinspired Optimization Methods and
                   their Application (BIOMA)}
}
012 2006-10-10 Thomas Weise BIOMA 2006 Presentation The presentation used to visualiza paper 013 at the BIOMA 2006 - The 2nd International Conference on Bioinspired Optimization Methods and their Application in Ljubljana (Slovenia) at the Jožef Stefan International Postgraduate School on October the 10th, 2006.
011 2006-09-25 Stefan Niemczyk Performance Report With this (German-language) project final report, Stefan Niemczyk sums up his project work. He did a performance study comparing the DGPF's ability in numerical regression to another framework. Although the other (c-language) framework is faster in time since the DGPF supports a lot of hooks and investigation/statistical utilities not needed for numerical regression, the evolutionary strategies of the DGPF allowed us to solve the tasks with less individual evaluations often. Also, an example Sudoku-solver is introduced to demonstrate how easily the DGPF can be adapted for a new problem space.
010 2006-09-25 Marc Kirchhoff Simulated Annealing Implementation Report With this (German-language) project final report, Marc Kirchhoff sums up his project work. He implemented the Simulated Annealing Algorithm for the DGPF. Furthermore, this implementation is also able to interact with other optimization algorithms like GA or hill climbing, creating heterogeneous searches.
009 2006-09-25 Mirko Dietrich, Lado Kumsiashvilli, Alexander Podlich Management GUI Report With this (German-language) project final report, Mirko Dietrich, Lado Kumsiashvilli, and Alexander Podlich sum up their project work. They have created a management gui for the DGPF which allows to control clusters running the search algorithms in a distributed fashion. The gui allows you also to evaluate the searches statistically, including the drawing of fancy diagrams. See also Document 008.
008 2006-09-08 Alexander Podlich GUI Slides (1) With this set of (German-language) slides, Alexander Podlich presented his project work. He is one of three developers who created the graphical user interface for the DGPF. While he is responsible for the graphical front end, his two team members provided the communication backend of the interface (which will be presented in november). The result of their excellent work was a versatile control center which allows to view the current state of distributed search, their progress over time (using shiny diagrams) and to modify their parameters on runtime. See also Document 009.
007 2006-07-24 Thomas Weise Student Project Work This preliminary and short document defines some of the project works for students joining the DGPF-project. The document is written in german language since the projects will run at University of Kassel, Hesse, Germany. The document includes the requirement definition for the enhancement a management system for the DGPF-project, a cluster-deployment project, a project in which the tasks of the Web Service Challenge is to be solved genetically as well as the research topic "feedback-loops" for the DGPF parameter updating mechanism. This is the second season where we're able to provide projects for students, for the projects of the first season see document 003.
006 2006-07-17 Thomas Weise, Kurt Geihs KuVS 2006 Paper The paper presented in 5. GI/ITG KuVS Fachgespräch "Drahtlose Sensornetze" in Stuttgart (using presentation 005).
@proceedings{WG2006DGPFb,
  author        = {Thomas Weise, Kurt Geihs},
  title         = {Genetic Programming Techniques for Sensor Networks},
  month         = {Jul},
  number        = {5},
  year          = {2006},
  abstract      = {In this paper we present an approach to automated program code generation
                   for sensor nodes and other small devices. Using Genetic Programming, we
                   are able to discover algorithms that solve certain problems. Furthermore,
                   non-functional properties like code size, memory usage, and communication
                   frequency can be optimized using multiobjective search techniques. The
                   evolution of algorithms requires program testing, which we perform using
                   a customized simulation environment for sensor networks. The simulation
                   model takes into account characteristic features of sensor nodes, such as
                   unreliable communication and resource constraints. An application example
                   is presented that demonstrates the feasibility of our approach and its
                   potential to create robust and adaptive code for sensor network
                   applications.},
  keywords      = {Genetic Programming, Genetic Algorithms, DGPF, Sensor Network, Sensor Node},
  URL           = {http://dgpf.sourceforge.net/documents/006-2006-07-17-kuvs_paper.pdf},
  pages         = {21-25},
  type          = {Resaerch Talk Paper},
  affiliation   = {University of Kassel,
                  Distributed Systems Group},
  location      = {Universit of Stuttgart},
  language      = {en},
  rights        = {unrestricted},
  booktitle     = {Proceedings of 5. GI/ITG KuVS Fachgespräch \"Drahtlose Sensornetze\"}
}
005 2006-07-17 Thomas Weise KuVS 2006 Presentation This is the (german tongue) presentation which I held at 5. GI/ITG KuVS Fachgespräch "Drahtlose Sensornetze" in Stuttgart. It shortly summarizes what inspired our work with GP and wireless sensor networks, what we already did and what we plan to achieve in near future. See paper 006.
004 2006-04-20 Thomas Weise Project Internals This set of slides give an introduction to the internals of the DGPF project. We describe the common Search API, the Genetic Algorithms Layer and its four distribution methods (local, client/server, peer-to-peer, peer-to-peer + client/server hybrid). An introduction is given to the Genetic Programming Layer, like it was done in document 002.
003 2006-03-10 Thomas Weise Student Project Work The distribution descriptions of this presentation have been outdated by the release 3.0.0 (2006-04-14) of the Framework. This short document defines some of the project works for students joining the DGPF-project. The document is written in german language since the projects will run at University of Kassel, Hesse, Germany. The document includes the requirement definition for a management system (including a GUI) for the DGPF-project as well as the research topic "feedback-loops" for the DGPF parameter updating mechanism.
002 2006-02-17 Thomas Weise Presentation for Interested Students The distribution descriptions of this presentation have been outdated by the release 2.0.0 (2006-03-02) of the Framework. The presentation includes descriptions about the client-server distribution of the computation load, which has been replaced by a peer-to-peer architecture then (and re-introduced later). This presentation was held for students that are interested in participating in the DGPF project. It shortly outlines the projects characteristics and gives some examples on how someone can contribute.
001 2006-01-20 Thomas Weise DGPF Paper The distribution descriptions of this paper have been outdated by the release 2.0.0 (2006-03-02) of the Framework. The presentation includes descriptions about the client-server distribution of the computation load, which has been replaced by a peer-to-peer architecture then (and re-introduced later). This paper explains some of the basic ideas behind the DGPF.
@unpublished{W2006DGPFa,
  author        = {Thomas Weise},
  title         = {Genetic Programming for Sensor Networks},
  month         = {Jan},
  year          = {2006},
  abstract      = {In this paper we present an approach to automated program code generation
                   for sensor nodes and other small devices using Genetic Programming. We give
                   a short introduction to Genetic Algorithms. Our new Distributed Genetic
                   Programming Framework facilitates the development of sensor network
                   applications. Genetic evolution of programs requires program testing.
                   Therefore we use a simulation environment for distributed systems of sensor
                   nodes. The simulation model takes into account characteristic features of
                   sensor nodes, such as unreliable communication and resource constraints.
                   Two application examples are presented that demonstrate the feasibility of
                   our approach and its potential to create robust and adaptive code for sensor
                   network applications.},
  keywords      = {Genetic Programming, Genetic Algorithms, DGPF, Sensor Network, Sensor Node},
  URL           = {http://dgpf.sourceforge.net/documents/001-2006-01-20-paper.pdf},
  affiliation   = {University of Kassel,
                  Distributed Systems Group},
  language      = {en},
  rights        = {unrestricted},
  }

statistics online since 2006-01-02.   RSS Feed
Contact us by sending an email to tweise@gmx.de to receive further information, to report errors, or to join our project.
All content on this site (http://dgpf.sourceforge.net/) is LGPL-licensed.
http://dgpf.sourceforge.net/documents/index.php last modified at 2015-07-22 04:10:52 GMT+0000 served at 2017-12-13 22:17:26 GMT+0000.
Valid CSS Valid XHTML 1.1
Valid RSS SourceForge.net Logo