Erik Goodman

Followers

Following

Co-authors

Public Views

Erik D. Goodman is PI and Director of the BEACON Center for the Study of Evolution in Action, an NSF Science and Technology Center headquartered at Michigan State University, funded at $47.5 million for 2010-20, with a diverse research program and extensive education and outreach programs. BEACON now has a very diverse membership of over 600, including evolutionary biologists and computer scientists/engineers studying evolutionary computation or digital evolution. Goodman received the Ph.D., computer and communication sciences, University of Michigan, 1972. He joined MSU’s faculty in Electrical Engineering and Systems Science in 1971, was promoted to full professor in 1984, and also holds appointments in Mechanical Engineering and in Computer Science and Engineering, in which he has guided many Ph.D. students. He directed the Case Center for Computer-Aided Engineering and Manufacturing from 1983-2002, and founded and directed MSU’s Manufacturing Research Consortium from 1993-2003. He co-founded MSU’s Genetic Algorithms Research and Applications Group (GARAGe) in 1993, conducting many projects under industrial sponsorship. In 1999, he co-founded Red Cedar Technology, Inc., (now a subsidiary of Siemens) which develops design optimization software, and was Vice President for Technology until BEACON was founded in 2010. He was chosen Michigan Distinguished Professor of the Year, 2009, by the Presidents Council, State Universities of Michigan. He was given MSU’s Distinguished Faculty Award in 2011. He was Chair of the Executive Board and a Senior Fellow of the International Society for Genetic and Evolutionary Computation, 2003-2005, and was the founding chair of ACM’s SIG on Genetic and Evolutionary Computation (SIGEVO) in 2005. He also co-leads an information and communication technology for development project in Tanzania.

less

Interests

Uploads

Papers by Erik Goodman

Introduction to genetic algorithms

Modeling genetic algorithm dynamics for onemax and deceptive functions

In this dissertation, we develop a model predicting dynamics of the counting-ones (OneMax) and a ... more In this dissertation, we develop a model predicting dynamics of the counting-ones (OneMax) and a form of deceptive function problems. The model describes the mean allele and, in the case of deceptive function, mean deceptive block values. The genetic algorithm (GA) that is being modeled consists of two-point crossover, fitness proportional selection and mutation operators. The model is developed to estimate the average GA dynamics, but it can be used for an individual run of the GA. First, we develop the model for the OneMax problem with very high crossover rates. Then, we modify the model by using statistics of very early generations from GA runs, to describe the complete dynamics for different (lower) crossover rates of the OneMax problem. In the development of the model, we introduce a new quantity that measures the effect of the crossover operation and is independent of generation, for practical purposes. Then, the model is generalized to cover other cases of the OneMax, such as weighted OneMax, as well as a form of deceptive function problem, for high enough crossover rates. The model is also modified to include Boltzmann selection with fixed or scaled selection pressures. The model can be applied to OneMax and deceptive function problems even when the crossover, mutation and the selection pressures (in the case of Boltzmann scaling) are changed at predetermined generations during a GA run. Since our model estimates the mean value (and, mean deceptive block value for deceptive function) at each locus at any generation, it can be used to determine a suitable migration time as well as the migration rate for parallel genetic algorithms (in the island model case) when migrations are allowed at any generation for our benchmark problems. Although the problems for which our model proved successful were rather simple or idealized, they were often sufficiently involved to capture interesting nontrivial features of the GA dynamics. The author hopes to extend the approach to model solution of more representative real-world problems with various degrees of OneMax similarity and various amounts of deception. At the end of the dissertation, an attempt to develop a stochastic differential equation model to predict the evolution of the fitness distribution for the OneMax problem is also presented. Although our attempt was not successful, it shows a strong connection between fitness evolution and certain diffusion equations and points toward the possibility of the existence of a diffusion-type equation that could describe the OneMax and maybe other type of GA dynamics.

Towards an epigenetics-inspired control system for power dispatch problem

Proceedings of the Genetic and Evolutionary Computation Conference Companion, Jul 15, 2017

Artificial neural networks and genetic and evolutionary optimization algorithms are abstractions ... more Artificial neural networks and genetic and evolutionary optimization algorithms are abstractions of observed natural processes and are found to have a widespread applicability in solving difficult problems. Epigenetics is a biological mechanism in which heritable changes are passed from one generation to another without genetic alterations. These changes allow organisms to adapt to abrupt changes in the environment. We propose and present an epigenetics-inspired learning paradigm illustrated in a power dispatch problem in which generators may fail abruptly. The proposed system is trained on optimized data, and should be able to handle a wider variety of operating conditions. The proposed idea is generic and novel but requires further application to more complex real-world control system problems to fully ascertain its value.

Download

Adaptive behavior of simulated bacterial cells to nutritional shifts : technical report

Download

About the Orthogonal Parameterization of Sculptured Part Surfaces and Initial Tool Surfaces

Journal of Manufacturing Science and Engineering-transactions of The Asme, Nov 1, 1997

In the domain of multi-axis NC machining of sculptured surface parts, the use of orthogonal param... more In the domain of multi-axis NC machining of sculptured surface parts, the use of orthogonal parameterizations of part and tool surfaces is convenient because it simplifies the transformation of coordinate systems. Using the so-called “differential-geometric method of sculptured surface NC machining,” developed by one of the authors, many parameterizations of part and tool surfaces are easily shown not to be orthogonal. To transform nonorthogonal part and tool surface parameterizations into orthogonal ones, the Jacobian of the transformation may be used. In cases when the Jacobian of the transformation is not known, it is possible to use differential equations for isogonal trajectories on the surfaces (choosing an orthogonal case), or a special kinematic method for obtaining sculptured surface equations. Influences of coordinate system transformations (translations and rotations along and about axes through the origin) on example part and tool surface parameterizations for four types of general helicoidal surfaces are described. The results mentioned above simplify the analytical description of the multi-axis NC machining process, and may be useful for writing NC toolpath generation software.

Srde

In this paper, we propose a methodology to improve the performance of the standard Differential E... more In this paper, we propose a methodology to improve the performance of the standard Differential Evolution (DE) in constraint optimization applications, in terms of accelerating its search speed, and improving the success rate. One critical mechanism embedded in the approach is applying Stochastic Ranking (SR) to rank the whole population of individuals with both objective value and constraint violation to be compared. The ranked population is then in a better shape to provide useful information e.g. direction to guide the search process. The performance of the proposed approach, which we call SRDE (Stochastic Ranking based Differential Evolution) is investigated and compared with standard DE with two variants of mutation strategies. The experimental results show that SRDE outperforms, or at least is comparable with standard DE in both variants in all the tested benchmark functions.

Download

Introduction to genetic algorithms

Find loads of the an introduction to genetic algorithms book catalogues in this site as the choic... more Find loads of the an introduction to genetic algorithms book catalogues in this site as the choice of you visiting this page. You can also join to the website book library that will show you numerous books from any types. Literature, science, politics, and many more catalogues are presented to offer you the best book to find. The book that really makes you feels satisfied. Or that's the book that will save you from your job deadline.

MOEA/D with Angle-based Constrained Dominance Principle for Constrained Multi-objective Optimization Problems

arXiv (Cornell University), Feb 10, 2018

This paper proposes a novel constraint-handling mechanism named angle-based constrained dominance... more This paper proposes a novel constraint-handling mechanism named angle-based constrained dominance principle (ACDP) embedded in a decomposition-based multi-objective evolutionary algorithm (MOEA/D) to solve constrained multiobjective optimization problems (CMOPs). To maintain the diversity of the working population, ACDP utilizes the information of the angle of solutions to adjust the dominance relation of solutions during the evolutionary process. This paper uses 14 benchmark instances to evaluate the performance of the MOEA/D with ACDP (MOEA/D-ACDP). Additionally, an engineering optimization problem (which is I-beam optimization problem) is optimized. The proposed MOEA/D-ACDP, and four other decomposition-based CMOEAs, including C-MOEA/D, MOEA/D-CDP, MOEA/D-Epsilon and MOEA/D-SR are tested by the above benchmarks and the engineering application. The experimental results manifest that MOEA/D-ACDP is significantly better than the other four CMOEAs on these test instances and the real-world case, which indicates that ACDP is more effective for solving CMOPs.

Download

Proceedings of the first ACM/SIGEVO Summit on Genetic and Evolutionary Computation

These proceedings contain the papers presented at the 2009 World Summit on Genetic and Evolutiona... more These proceedings contain the papers presented at the 2009 World Summit on Genetic and Evolutionary Computation (2009 GEC Summit), held in Shanghai, China, June 12-14, 2009. 2009 GEC Summit is sponsored and organized by ACM SIGEVO, the Special Interest Group for Genetic and Evolutionary Computation, sponsor of the annual GECCO conferences. The 2009 GEC Summit will feature the latest research and demonstrated successes in this dynamic area, including new approaches and breakthrough applications to problems in biology, medicine, engineering design, agriculture, logistics, traffic, security, scheduling, military affairs, and other fields. It maintains an impressive record of both submission totals and acceptance rate. This year there were 372 papers submitted, with 153 accepted as full papers, giving an acceptance rate of 41%. This acceptance rate represents a healthy selection pressure in order to preserve the quality of the conference, and even though the proceedings are not bound by physical limits on the number of accepted papers, the organizers have decided to keep the acceptance rate at the lower end. The 2009 GEC Summit has adopted electronic proceedings, and we are confident that our attendees will be pleased with this format, as it greatly facilitates the keeping of all conference materials and automated searching for topics of interest. The proceedings will appear in the ACM Digital Library, making them available to the world. The 2009 GEC Summit used a strict paper review system. To reduce any bias that reviewers might have, all reviews were conducted double blind. No author's names were included in the reviewed papers. Each paper had at least 4 reviewers while each reviewer had at least 5 papers to review. About 403 researchers participated in the review process. Their work is much appreciated and is absolutely vital for the quality of the conference. In addition to the presentation of the papers contained in these proceedings, the 2009 GEC Summit also includes free tutorials. Registered participants may attend any of the 100-minute tutorials, presented by some of the world's leading experts in evolutionary computation. Tutorials are distributed throughout the three days of the conference, allowing each participant to decide, at most time slots, to attend either a tutorial or one of several sessions in which accepted papers are presented. Tutorials are grouped into introductory level, advanced, and specialized tutorials. In general, introductory tutorials in a given area will be scheduled ahead of the advanced tutorials, allowing the interested participant to attend them both in order.

Predicting Leaf Expansion of ‘Montmorency’ Sour Cherry from Degree-day Accumulations1

Journal of the American Society for Horticultural Science, Sep 1, 1982

Second-order rectangular hyperbolic and asymptotic models which predict spur and terminal leaf ex... more Second-order rectangular hyperbolic and asymptotic models which predict spur and terminal leaf expansion of 'Montmorency' sour cherry (Prunus cerasus L.) were developed from leaf area observations and temperature records made in orchards near East Lansing, Mich. Average leaf area per leaf was more highly correlated with degreeday accumulation at a base of 4°C starting April 19, than with day of the year. Leaf area per leaf increased linearly with degree-day accumulation until full leaf expansion. Final spur or terminal leaf size was not constant between years.

Download

Genetic Algorithm-Based Design of Energy Absorbing Laminated Composite Beams

This paper describes a general approach to structural design using Genetic Algorithms, and an app... more This paper describes a general approach to structural design using Genetic Algorithms, and an application of that approach to the design of energy absorbing laminated composite beams containing distributed thin, compliant layers. We first discuss a method for applying a Genetic Algorithm (GA) to structural design, using it as an evolutionary search optimizer in conjunction with a structural simulator as its objective function. The simulator used is an efficient and robust special purpose finite element model based on a layerwise laminate theory. The GA “designs” the beam by selecting material assignments for the subregions and the locations of compliant layers, and evaluates the design using the simulator. The efficiency of the GA search is improved by use of the “injection island” architecture. The results demonstrate that the parallel GA architectures achieved algorithmic superlinear speedup to similar quality of solution in comparison with single-population genetic algorithms.

A concurrent error detection and correction algorithm for fault-tolerant vlsi arithmetic array processors

A concurrent error detection and correction algorithm for errors caused by permanent, intermitten... more A concurrent error detection and correction algorithm for errors caused by permanent, intermittent and transient faults in arithmetic parallel-pipeline array processors, is described. The fault model, applicable to VLSI implementations, assumes the occurrence of faults with unknown origin and frequency. Recovery from faults is achieved through minimal processing element (PE) redundancy in the array architecture, facilitated by spatial bypassing of the correct operands previous to the occurrence of a fault to fault-free PE's and recomputing during the following clock cycles. The overhead hardware and timing are determined. It is shown that this concurrent error detection and correction technique uses less additional hardware than RESO (REcomputing with Shifted Operands), offers marginal timing improvement over the RESO technique, and adds the concurrent error correction capability which is not present in RESO. Furthermore, the additional hardware and control for error detection and recovery is local and modular, hence making this technique very attractive for VLSI implementations. This algorithm is based on a general fault model which assumes the occurrence of permanent, intermittent and transient fault of unknown origin and frequency. Transient and intermittent faults with a duration longer than one computational clock cycle are classified as "permanent". The effects of these faults are detected as undesired changes in the logical values at the outputs of the PE's or other circuit elements, such as latches. The error detecting algorithm is concurrent and thus may detect and correct errors caused by transient faults affecting the PE's randomly.

Control allocation-based adaptive control for greenhouse climate

International Journal of Systems Science, Feb 26, 2018

This paper presents an adaptive approach to greenhouse climate control, as part of an integrated ... more This paper presents an adaptive approach to greenhouse climate control, as part of an integrated control and management system for greenhouse production. In this approach, an adaptive control algorithm is first derived to guarantee the asymptotic convergence of the closed system with uncertainty, then using that control algorithm, a controller is designed to satisfy the demands for heat and mass fluxes to maintain inside temperature, humidity and CO 2 concentration at their desired values. Instead of applying the original adaptive control inputs directly, second, a control allocation technique is applied to distribute the demands of the heat and mass fluxes to the actuators by minimising tracking errors and energy consumption. To find an energy-saving solution, both single-objective optimisation (SOO) and multiobjective optimisation (MOO) in the control allocation structure are considered. The advantage of the proposed approach is that it does not require any a priori knowledge of the uncertainty bounds, and the simulation results illustrate the effectiveness of the proposed control scheme. It also indicates that MOO saves more energy in the control process.

Download

A hands-on paradigm for EAP education: undergraduates, pre-college students, and beyond

Proceedings of SPIE, Apr 6, 2007

Electroactive polymers (EAPs) are receiving increasing interest from researchers due to their uni... more Electroactive polymers (EAPs) are receiving increasing interest from researchers due to their unique capabilities and numerous potential applications in biomimetic robots, smart structures, biomedical devices, and micro/nanomanipulation. Since these materials are relatively new, it is imperative to educate students and the general public to raise their awareness of EAP potentials and produce the talent pool needed for continuing, rapid advances in the field of EAPs. In this paper we describe our concerted effort in teaching EAP to undergraduates, grade school students, and the general public, through hands-on research and learning on EAP-based biomimetic robots. Two integrated activities are highlighted: A senior Capstone design program on EAP robots, and the subsequent programs that use these developed robots to reach out to pre-college students. A robotic fish and a sociable robot enabled by ionic polymer-metal composite materials are used as examples throughout the paper.

Download

Multi-layer hierarchical optimisation of greenhouse climate setpoints for energy conservation and improvement of crop yield

Biosystems Engineering, May 1, 2021

A collaboration-based particle swarm optimizer with history-guided estimation for optimization in dynamic environments

Expert Systems With Applications, Apr 1, 2019

This paper introduces a collaboration-based particle swarm optimizer (PSO) by incorporating three... more This paper introduces a collaboration-based particle swarm optimizer (PSO) by incorporating three new strategies: a global learning strategy, a probability of learning, and a "worst replacement" swarm update rule. Instead of learning from the personal historical best position and the global (or local) best position which was used by the classical PSO, a target particle learns from another randomly chosen particle and the global best one in the swarm. Instead of accepting a new velocity directly, the velocity updates according to a learning probability, according to which the velocity of the target particle in each dimension updates via learning from other particles or simply inherits its previous velocity component. Since each particle has the same chance to be selected as a leader, the worst particle might influence the whole swarm's performance. Therefore, the worst particle in the swarm in each update is moved to a new better position generated from another particle. The proposed algorithm is shown to be statistically significantly better than six other state-of-the-art PSO variants on 20 typical benchmark functions with three different dimensionalities.

Download

Nearly dynamic programming NN‐approximation–based optimal control for greenhouse climate: A simulation study

Optimal Control Applications & Methods, Oct 12, 2017

Greenhouse production must create a suitable growth environment for its crop to improve yield whi... more Greenhouse production must create a suitable growth environment for its crop to improve yield while minimizing the energy used to maintain the greenhouse environment, thereby reducing production cost. To this end, this paper develops a nearly optimal control approach based on adaptive dynamic programming .In this method, three neural networks(NN) are used to estimate the value function and control policy, and to compensate for the unmodeled dynamics of the greenhouse climate. Taking into account the greenhouse system typically being an over-actuated system, the total control efforts for heat, fog and CO 2 are considered as the virtual control inputs to be generated by the optimal controller. To obtain the real control inputs, a control allocation technique is introduced to distribute the virtual control inputs to the actuators. Finally, Lyapunov stability analysis is performed to derive the update law of the NNs to ensure the asymptotic convergence of the closed-loop system, and simulation is carried out to illustrate the effectiveness and control performance of the proposed approach.

Download

Optimal design of laminated composite structures using coarse-grain parallel genetic algorithms

Computing Systems in Engineering, Aug 1, 1994

A new coarse-grain parallel architecture for genetic algorithms, called island injection genetic ... more

Evolutionary multi-objective automatic clustering enhanced with quality metrics and ensemble strategy

Knowledge Based Systems, 2020

This is a PDF file of an article that has undergone enhancements after acceptance, such as the ad... more This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Download

A neighbor-based learning particle swarm optimizer with short-term and long-term memory for dynamic optimization problems

Information Sciences, Jul 1, 2018

This paper presents a novel Particle Swarm Optimization algorithm to address Dynamic Optimization... more This paper presents a novel Particle Swarm Optimization algorithm to address Dynamic Optimization Problems. The algorithm incorporates a neighbor-based learning strategy into the velocity update of Particle Swarm Optimization, in order to enhance the exploration and exploitation capabilities of particles. Unlike the traditional swarm update scheme, a "worst replacement" strategy is used to update the swarm, whereby the position of the worst particle in the swarm is replaced by a better newly generated position. The shortterm memory is employed to store solutions with intermediate fitnesses from the most recent environment, and the long-term memory is to store the historical best solutions found in all previous environments. After an environmental change is detected, some particles' positions in the swarm are replaced by the members of the short-term memory, and the best member in the long-term memory under the current environment is re-introduced to the active swarm along with its Gaussian neighborhood, then the remaining particles' positions are re-initialized. The performance of the proposed algorithm is compared with six state-of-the-art dynamic algorithms over the Moving Peaks Benchmark problems and Dynamic Rotation Peak Benchmark Generator. Experimental results indicate that out algorithm obtains superior performance compared with the competitors.

Download

Erik Goodman

Uploads

Papers by Erik Goodman

Log In