Pareto Optimality in Coevolutionary Learning

Ficici, Sevan G.; Pollack, Jordan

doi:10.1007/3-540-44811-x_34

Cited by 80 publications

(75 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More difficult tests, on the other hand, are typically characterized by greater variability in the interaction outcomes, thus we expect SFIMX to commit greater errors in their estimation. In the second variant, Dist, we employ the concept of distinctions, borrowed from competitive coevolution [8]. A test t is said to make a distinction between programs p 1 and p 2 if g(p 1 , t) = g(p 2 , t).…”

Section: Extending Sfimx With Adaptive Test Selectionmentioning

confidence: 99%

“…A competitive coevolutionary algorithm, with a separate population of programs and a separate population of tests that coevolve with them, can be seen as an emerging way of adjusting those probabilities. The extent to which this analogy holds depends on several factors, including the way in which the tests are evaluated (note the remark in Section 3 on SFIMX-Dist being inspired by the concept of distinctions, a specific way of evaluating tests in competitive coevolution [8]). A range of works in GP investigated the possibility of coevolving tests, starting from the seminal work by Pagie and Hogeweg [28], to the idea of coevolving fitness predictors along with programs [30].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Adaptive Test Selection for Factorization-based Surrogate Fitness in Genetic Programming

Krawiec

Liskowski

2017

Foundations of Computing and Decision Sciences

View full text Add to dashboard Cite

Abstract. Genetic programming (GP) is a variant of evolutionary algorithm where the entities undergoing simulated evolution are computer programs. A fitness function in GP is usually based on a set of tests, each of which defines the desired output a correct program should return for an exemplary input. The outcomes of interactions between programs and tests in GP can be represented as an interaction matrix, with rows corresponding to programs in the current population and columns corresponding to tests. In previous work, we proposed SFIMX, a method that performs only a fraction of interactions and employs non-negative matrix factorization to estimate the outcomes of remaining ones, shortening GP's runtime. In this paper, we build upon that work and propose three extensions of SFIMX, in which the subset of tests drawn to perform interactions is selected with respect to test difficulty. The conducted experiment indicates that the proposed extensions surpass the original SFIMX on a suite of discrete GP benchmarks.

show abstract

Section: Extending Sfimx With Adaptive Test Selectionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Adaptive Test Selection for Factorization-based Surrogate Fitness in Genetic Programming

Krawiec

Liskowski

2017

Foundations of Computing and Decision Sciences

View full text Add to dashboard Cite

show abstract

“…The traditional evolutive computation techniques have several disadvantages. Coevolution has been proposed as a way to evolve a learner and a learning environment simultaneously such that open-ended progress arises naturally, via a competitive arms race, with minimal inductive bias [15]. The viability of an arms race relies on the sustained learnability [43] of environments.…”

Section: Coevolution In Roboticsmentioning

confidence: 99%

Untitled

Berlanga

Sanchís

Isasi

et al. 2002

Journal of Intelligent and Robotic Systems

View full text Add to dashboard Cite

Abstract. In this paper, a new coevolutive method, called Uniform Coevolution, is introduced to learn weights of a neural network controller in autonomous robots. An evolutionary strategy is used to learn high-performance reactive behavior for navigation and collisions avoidance. The introduction of coevolutive over evolutionary strategies allows evolving the environment, to learn a general behavior able to solve the problem in different environments. Using a traditional evolutionary strategy method, without coevolution, the learning process obtains a specialized behavior. All the behaviors obtained, with/without coevolution have been tested in a set of environments and the capability of generalization is shown for each learned behavior. A simulator based on a mini-robot Khepera has been used to learn each behavior. The results show that Uniform Coevolution obtains better generalized solutions to examples-based problems.

show abstract

“…Studies that have investigated this issue have proposed diversity maintenance techniques to introduce and maintain diversity in the population through specific use of selection and variation processes, which have been shown subsequently to improve the performance of coevolutionary learning [27], [29], [31], [32]. However, it is not known whether there is a relationship between generalization performance and diversity in coevolutionary learning.…”

mentioning

confidence: 99%

“…Establishing any direct relationship between generalization performance and diversity theoretically would be very difficult. Although earlier studies have turned to various empirical approaches, there is a lack of rigorous analysis to investigate this issue in terms of generalization performance measures in coevolutionary learning [29], [32], [33]. Furthermore, these past studies made little attempt to measure diversity levels in coevolutionary learning despite strong claims that they are increased while investigating their impact on performance [24], [25], [29], [33].…”

mentioning

confidence: 99%

Relationship Between Generalization and Diversity in Coevolutionary Learning

Chong

Tiňo

Yao

2009

IEEE Trans. Comput. Intell. AI Games

View full text Add to dashboard Cite

Abstract-Games have long played an important role in the development and understanding of coevolutionary learning systems. In particular, the search process in coevolutionary learning is guided by strategic interactions between solutions in the population, which can be naturally framed as game playing. We study two important issues in coevolutionary learning-generalization performance and diversity-using games. The first one is concerned with the coevolutionary learning of strategies with high generalization performance, that is, strategies that can outperform against a large number of test strategies (opponents) that may not have been seen during coevolution. The second one is concerned with diversity levels in the population that may lead to the search of strategies with poor generalization performance. It is not known if there is a relationship between generalization and diversity in coevolutionary learning. This paper investigates whether there is such a relationship in coevolutionary learning through a detailed empirical study. We systematically investigate the impact of various diversity maintenance approaches on the generalization performance of coevolutionary learning quantitatively using case studies. The problem of the iterated prisoner's dilemma (IPD) game is considered. Unlike past studies, we can measure both the generalization performance and the diversity level of the population of evolved strategies. Results from our case studies show that the introduction and maintenance of diversity do not necessarily lead to the coevolutionary learning of strategies with high generalization performance. However, if individual strategies can be combined (e.g., using a gating mechanism), there is the potential of exploiting diversity in coevolutionary learning to improve generalization performance. Specifically, when the introduction and maintenance of diversity lead to a speciated population during coevolution, where each specialist strategy is capable of outperforming different opponents, the population as a whole can have a significantly higher generalization performance compared to individual strategies.

show abstract

Pareto Optimality in Coevolutionary Learning

Cited by 80 publications

References 13 publications

Adaptive Test Selection for Factorization-based Surrogate Fitness in Genetic Programming

Adaptive Test Selection for Factorization-based Surrogate Fitness in Genetic Programming

Untitled

Relationship Between Generalization and Diversity in Coevolutionary Learning

Contact Info

Product

Resources

About