First International Workshop on Computational Synthesis of Systems from Building Blocks (CSSB2013) at GECCO 2013, July 6, 2013, Amsterdam, Netherland
Automated computational synthesis of circuit systems, mechatronic systems, control systems, optical systems, antennas, materials, and other structures by evolutionary algorithms such as genetic programming have attracted significant interest in the past decades. There is a core common element to these problems: a set of building blocks need to be selected, tuned and assembled into a system with a particular target behavior. Candidate systems are evaluated by a simulator such as Pspice, Matlab, Dymola, Modelica, NetSim, Ansys, or VASP.
Despite significant interest, we have not seen broad industrial application of computational synthesis of systems. This is due to at least two reasons: limited computational resources / scalability of the search algorithms, and questionable trustworthiness / reliability of the synthesized designs. But there is hope: it has become easier to access large computational resources (e.g. Amazon Web Services), and recent research has uncovered techniques for synthesizing trustworthy and reliable systems. It is desirable for the evolutionary computation community to develop techniques for industrially relevant synthesis of systems.
This workshop invites leading researchers and engineers interested in computational synthesis of systems to discuss realizing real-world applications. We welcome people from the different application domains. We are especially interested in papers on:
1) Techniques to exploit massive computational resources for computational synthesis of systems. This includes parallel evolutionary algorithm models and open-source softwares based on MPI, TCP, Hadoop, etc. It could be based on current popular GP package such as open-beagle, ECJ, or GPLab.
2) Techniques to improve the trustworthiness, reliability, or overall quality of search results. This may include the degree of open-endedness, how the building blocks library is constructed, how constraints are modeled, and how multiple objectives and constraints are handled.
3) Solving problems from different application domains such as computational synthesis of analog / mixed-signal circuits, custom digital circuits, mechatronic systems, controllers, materials, proteins, and so on. Novel commercial or open-source simulators that are used by design community. Suggestion of computational synthesis problems from building blocks.
4) Improvements in scalability and quality of results via novel algorithms for computational synthesis. This may include improvements in topology search, parameter search, fitness approximation, module discovery and reuse, and computationally intensive simulation.
5) Benchmark problems for computational synthesis of systems. We welcome engineering design community to pose challenging design synthesis problems.
8.30 Workshop Introduction
9.20 Open-ended computational synthesis of mechatronic systems
9.50 Computational Material Discovery using evolutionary algorithms
10.20 Coffee Break
10.40 Computational synthesis of Analog circuits
11.10 Computational synthesis of kinematic mechanisms
11.40 Discussion Panel on open-source computational synthesis platform, benchmark problems, domain synthesis problems
12.30 Wrap up and Conclusions
2:00 Distributed models of parallel evolutionary algorithms (GP/GA) based on MPI or TCP
2:30 Simulation software survey and integrating genetic programming package with simulators
3:00-5:00 Individual paper presentations
March 28th paper submissions
April 15th for notifications
April 25th for camera-ready.
How to Submit papers:
To submit your contribution, send your pdf by e-mail to Jianjun Hu at email@example.com with email subject “CSSB2013 submission”. Papers should not exceed the limit of 8 pages and must meet with deadline of the workshop
Please note that all contributions must abide ACM formatting rules because all contributions will be in the GECCO companion proceedings as well as in the ACM digital library. Failing to comply with the ACM formatting rules will result in exclusion from the proceedings. For formatting details, visit http://www.sigevo.org/gecco-2013/papers.html.
Jianjun Hu, Assistant professor of computer science at the University of South Carolina. He directs the Machine Learning and Evolution Laboratory. He received a Ph.D from Michigan State University in area of evolutionary algorithms. His research is focused on genetic programming, computational synthesis of mechatronic systems, data mining and bioinformatics. He has developed the GPBG plaftform for open-ended dynamic system design based on bond graph models. He has been able to evolve analog filters, printer mechanisms, vibration absorbers, MEMS systems. He has also developed techniques for scalable GP based on sustainable evolutionary search model and balanced structure and parameter search. He has more than 25 publications in the area of evolutionary system synthesis.
Trent McConaghy, Co-founder and Chief Technology Officer of Solido Design Automation. Solido provides variation-aware design tools for custom circuit designers at Nvidia, Qualcomm, and Huawei and other leading semiconductor companies. Trent was Co-founder and Chief Scientist of Analog Design Automation, which was acquired by Synopsys in 2004. Prior to that, he did intelligent systems research for the Canadian Department of National Defense. He received his Ph.D. from KU Leuven, Belgium in 2008. In his award-winning thesis, he applied genetic programming to synthesis of trustworthy-by-construction analog circuit topologies. He is author of the books “Variation-Aware Analog Structural Synthesis: A Computational Intelligence Approach” and “Variation-Aware Design of Custom Integrated Circuits: A Hands-On Field Guide.”
Hod Lipson, Associate Professor of Mechanical & Aerospace Engineering and Computing & Information Science at Cornell University in Ithaca, NY. He directs the Creative Machines Lab, which focuses on novel ways for automatic design, fabrication and adaptation of virtual and physical machines. He has led work in areas such as evolutionary robotics, multi-material functional rapid prototyping, machine self-replication and programmable self-assembly. Lipson received his Ph.D. from the Technion – Israel Institute of Technology in 1998, and continued to a postdoc at Brandeis University and MIT. His research focuses primarily on biologically-inspired approaches, as they bring new ideas to engineering and new engineering insights into biology. His group has developed techniques to evolve robust analog circuits based on winSpice simulator and kinematic mechanisms.
Artem R. Oganov, Professor of Dept. of Geosciences and Dept. of Physics and Astronomy at State University of New York at Stony Brook. He received his PhD in Crystallography from University College London. His research is in the fields of mineral physics, computational materials science and materials design. His contributions include the development of a novel methodology for predicting crystal structures, discovery of several new minerals and novel forms of elements and compounds, development of methods for studying energy landscapes of chemical systems, development of accurate pressure scales, and extensive studies of structural stability and properties of minerals and compounds. Professor Oganov is a founding member of the Mineralogical Crystallography group of the European Crystallographic Association, and a member of editorial boards of American Mineralogist and the Journal of Superhard Materials.
Andriy O. Lyakhov, Research Assistant Professor of Dept. of Geosciences at State University of New York at Stony Brook. He received his PhD in theoretical physics from Basel University, Switzerland. He is the Chief developer of USPEX code, a widely used open-source evolutionary crystal structure prediction tool written in Matlab. He has published more than 15 papers in the area of computational material discovery.
Accepted papers will appear in Workshop Proceedings of GECCO 2013 published by ACM! GECCO is a prestigious conference ranked as 11th best out of 701 considered conferences in “Artificial Intelligence / Machine Learning / Robotics / Human Computer Interaction.”
Contact organizers with questions: Jianjun Hu (firstname.lastname@example.org), Trent McConaghy(email@example.com)