BEACON is strongly interested in international research collaborations. While BEACON cannot export research funding abroad, BEACONites can participate in collaborative projects, with work in the U.S. funded by BEACON in collaboration with work funded from other sources in other countries. BEACON can also bring international researchers to BEACON for visits to further such programs.

Examples of BEACON’s international research projects 

New collaboration with the Center for Evolutionary Intelligence and Engineering Applications 

2014-07-15 10.16.14On July 12, 2014, BEACON established a collaborative relationship with a new research center just founded at Shantou University in southern China. Prof. Zhun Fan, who received his Ph.D. at Michigan State University, was named as the Director of the new Center for Evolutionary Intelligence and Engineering Applications. Prof. Fan and Erik Goodman, BEACON’s Director, have collaborated on design of mechatronic systems using bond graphs and genetic programming since Fan’s graduate student days at MSU and during his time on the faculty of the Technical University of Denmark.

The collaboration will center on development of new design techniques to leverage the power of genetic programming for design of mechatronic systems (including MEMS, or micro-electro-mechanical systems). Videoconference and frequent visits by team members in both directions are expected. 

BEACON researchers collaborate with New Zealanders to model, apply ideas from research on epigenetics

From left: Oliver Chikumbo, Erik Goodman, Kalyanmoy Deb

Dr. Oliver Chikumbo, of Scion, a Crown Research Institute in Rotorua, New Zealand, has been collaborating with BEACON Profs. Erik Goodman and Kalyanmoy Deb, with Dr. Chikumbo having spent one month in 2011 and one month in 2013 as a BEACON visitor. The project arose from earlier collaborations between Goodman and Chikumbo beginning in the 1990’s and continuing at a low level ever since. The work is continuing, and the team has been joined by Daniel Couvertier, a BEACON graduate student in CSE at MSU, and Mr. Hyungon Kim, a graduate student in the Human Interface Technology Lab at the University of Canterbury (New Zealand), supervised by Prof. Gun Lee.

Pareto front visualisation after 100 generations: top left – Pareto points; top right – Grid plot; bottom left – Surface and contour plot; bottom right – Pareto points super-imposed on a grid.

Inspired by research on epigenetic mechanisms being done under the direction of Sir Peter Gluckman at the Liggins Institute in Auckland, Chikumbo suggested to Deb and Goodman that they apply some of these epigenetic ideas to their evolutionary computation approach for solving land use problems, which typically involve multi-objective optimization with dozens of options each year over a 10-50-year timeframe for hundreds or thousands of plots. The project depends heavily on being able to optimize a 14-objective problem, determining a Pareto set of optimal tradeoff solutions. The Evolutionary Multi-Objective Optimization (EMOO) search technology developed by the team complemented Prof. Deb’s R-NSGA-II algorithm with new epigenetic operators developed by the team and to be further studied by Daniel Couvertier. The size of the search space of potential solutions is on the order of 10**600, and with 14 objectives to evaluate, appears at first glance to defy any efforts at optimization. However, the combination of heuristics and decision-making processes used by the team (nicknamed “WISDOM”) was able to find a useful Pareto set of solutions that bore up well under critical examination. The team is also working with the University of Canterbury (in New Zealand) to develop virtual reality tools to help users comprehend the set of Pareto-optimal solutions. The decision-making approach given a set of optimal solutions involves allowing individual stakeholders to first express their relative preferences among the 14 objectives, and then to rank four solutions (selected according to their preferences from among the optimal set). These ranks are then combined using a scheme called the Analytical Hierarchy Process (AHP) to identify the solution most compatible with the preferences of all stakeholders.

Whole farm totals of all the 14 outputs over a period of 50 years from one of the Pareto points (-18.14 -85.97 13.25), from the valley, after 100 generations. Click to enlarge.

Whole farm totals of all the 14 outputs over a period of 50 years from one of the Pareto points (-18.14 -85.97 13.25), from the valley, after 100 generations. Click to enlarge.

The WISDOM process is applicable to many “wicked” societal problems, and allows stakeholders to address simultaneously economic, environmental and social concerns, to satisfy the “Triple Bottom Line.” The team plans to redevelop the platform for integrating the many distinct simulators used to calculate the 14 objective values for each solution into a sophisticated sensor and model integration framework in order to make it easier to generalize the approach for application to many problem domains, in partnership with a company with which negotiations are currently underway.

For this work, the team has been awarded the 2013 Wiley Practice Prize, a prestigious competition held every 2 years at the Multi-Criteria Decision Making Conference.

China’s Greenhouses of the Future—Evolutionary Controls to Lower Environmental/Economic Costs

China has ambitious plans to design and build a new generation of greenhouses, helping to supply its year-round needs for fresh vegetables in a way that is economically viable and environmentally friendly. Several projects have been funded under the leadership of Prof. Lihong Xu, by the National Natural Science Foundation of China and under China’s Twelfth Five-Year Plan, which now includes this BEACON collaboration. Prof. Xu spends a part of each year in residence at BEACON, and several of his students have also been visiting scholars in BEACON, working with Prof. Erik Goodman, BEACON’s Director. BEACON graduate student Jose Llera-Ortiz and BEACON postdoctoral researcher Prakarn Unachak are now working with Tongji University graduate student Chenwen Zhu (a visiting scholar in BEACON) to develop a sophisticated model of the greenhouse internal environment and a crop model, which are both needed to enable implementing the “Multi-Objective Compatible Controller,” the evolutionary methodology that the team has been developing. A new greenhouse is nearly complete on the campus of Tongji University, in which the new control methodology can be parameterized and tested. The team is collaborating on the plant physiology and greenhouse modeling with Prof. Erik Runkle at MSU and Prof. Weihong Luo (Nanjing Agricultural University, China), both of whom are experts in plant physiology and greenhouse modeling.

Evolutionary Algorithms for Enhanced Ultra-Wideband Microwave Imaging of Breast Cancer Tumors

BEACON Electrical Engineering Ph.D. student and NSF fellowship winner Blair Fleet

Under the leadership of Prof. Meng Yao, of East China Normal University (Shanghai, China), who is visiting in BEACON for part of each year under funding from his grants, a BEACON team, including Profs. Jack Deller and Erik Goodman and graduate student Blair Fleet, is working on generating images from the microwave data collected by Prof. Yao’s team at ECNU and Shanghai’s First People’s Hospital. Microwave imaging of tumors in breast tissue is difficult because microwaves do not normally penetrate deep into the tissue and reflect at levels that are easily resolved. However, a novel antenna design that matches the impedance at the skin boundary enables better penetration, and a frequency-stepping pulse sequence allows the depth of reflections due to changes in dielectric constant to be determined. The signal-to-noise ratio is low, however, so the plan is to use evolutionary algorithms to find suitable classifiers to enhance the tumor images in a way that is useful for diagnosis. BEACON Ph.D. student Blair Fleet, who won an NSF Graduate Fellowship to continue this work, is first constructing models of the signal propagation and reflection in the breast tissue, then will use those models for initial exploration of possible schemes for distinguishing cancerous tissue from normal tissue and tissue containing abnormal but non-cancerous structures. Two other BEACON graduate students are working as part of this team: Jinyao Yan and Pedro Nariyoshi, both of whom are tackling signal processing problems related to the project.

Comments are closed.