141 research outputs found
The secondary substrate problem in Co-Evolution and Developmental-Evolution
No full textThe performance of an Evolutionary Algorithm on a search problem is critically effected by the substrate used to encode the candidate solutions of the problem. In addition to the challenge of designing evolvable genetic substrates, two-population competitive coevolutionary algorithms (coEAs) and developmental Evolutionary Algorithms(devo-EAs) present another substrate-related design problem. Both involve an additional substrate with its own mechanism of change. In coEAs, test-cases are encoded with an independent genetic substrate having its own variation operators. In devo-EAs, phenotypes are composed of a distinct substrate with associated generative mechanisms capable of changing an individual's form and size during development. Though this "secondary" substrate is a distinctive feature of both algorithms, the design problem it poses remains poorly understood. This dissertation proposes novel formal models to characterize how the properties of the secondary substrate influences the performance devo-EAs and coEAs respectively.
Firstly, we propose a computational model for devo-EAs which shows that the point in time at which the development of a phenotype halts can introduce selection biases that can cause an empirically measurable retardation in the performance of a devo-EA. Furthermore, a Genotype-Phenotype map that is bias-free is formally equivalent to a Nash equilibrium in a non-cooperative multi-player game, where each genotype is a player, the possible halting points are strategies and the payoffs are related to the fitness function. We show that algorithmic solutions to find this Nash map are expensive without a suitable secondary substrate.
Secondly, we propose a novel search space model for Pareto coevolution that formally defines the evolvability properties required of the secondary substrate for pathology-free learning with a mutation-only coEA.With this model, we show that on boolean classification problems (a) the variational properties of the secondary substrate are a property of the problem class rather than tied to individual problems, and (b) the absence of coevolutionary pathologies does not imply success in finding high-quality solutions. Rather than being mysterious dynamical properties of coEAs, these findings are transparently explained using Machine Learning first principles
How artificial ontogenies can retard evolution
No full textRecently there has been much interest in the role of indirect genetic encodings as a means to achieve increased evolvability. From this perspective, artificial ontogenies have largely been seen as a vehicle to relate the indirect encodings to complex phenotypes. However, the introduction of a development phase does not come without other consequences. We show that the conjunction of the latent ontogenic stucture and the common practice of only evaluating the final phenotype obtained from development can have a net retarding effect on evolution. Using a formal model of development, we show that this effect arises primarily due to the relation between the ontogenic structure to the fitness function, which in turn impacts the properties being evaluated and selected for during evolution. This effect is empirically demonstrated with a toy search problem using LOGO-turtle based embryogenic processes
Product Configuration Optimization For Disassembly Planning: A Differential Approach
No full textProduct configuration immensely influences the suitability of a product for end-of-life (EOL) disassembly. The product configuration is the relative spatial and logical arrangement of the different parts/sub-assemblies of the product with respect to each other. The complexity involved in studying the influence of configuration design on EOL disassembly has limited the scope of the current design for disassembly (DfD) approaches to guideline-based prescriptive methods and index-based evaluation techniques. The application of these approaches has primarily been limited to specific case studies of product redesign. Many of the current methods do not provide the necessary rigor that will lead to the creation of a theoretical base for addressing product configuration issues which is indispensable during product redesign. Though fraught with obstacles, studying the effects of product configuration on DfD will be useful to develop automated configuration optimization methods for EOL disassembly. To this end, a model to study the combinatorial configuration design optimization problem from a disassembly perspective is described in this study. The different structural principles of the design space derived in this study provide insights into the possibilities and the natural shortcomings of automated optimization of a product by relating the effects of design constraints and disassembly requirements on product redesign. A hierarchical evolutionary programming based algorithm is also developed to test the design solutions generated by the proposed model
Disassembly-Oriented Product Classification using Neural Networks
No full textThe increasing importance of a product\u27s relationship and effects on the environment has prompted active research in Environment Conscious Design and Manufacture (ECDM). Disassembly of a product makes a product\u27s parts available for recycling, remanufacturing or reuse. With the increasing complexity and variety of new products, the emergence of dedicated facilities to handle the enormity of this disassembly task in the near future is forseeable. Grouping diverse products based on their similarity in dissassembly characteristics would result in a more effective and flexible use of the capabilities of different disassembly factories. A modification of the growing neural gas network model was found to be effective to implement this grouping
Towards an Evolutionary-Developmental Approach for Real-World Substrates
No full textExtending "body-brain" evolution to the real-world presents a number of difficulties due to conflicting idealizations between evolutionary and constructional models. Toward addressing this gap, we develop a simple model system to analyze the effects of undoing these idealizations. Preliminary experiments with this system show that high variability developmental substrates can influence evolutionary dynamics by causing ambiguities in selection. Furthermore the substrate can enable the evolution of adaptive responses to nondeterministic developmental effects
How Artificial Ontogenies Can Retard Evolution
No full textRecently there has been much interest in the role of indirect genetic encodings as a means to achieve increased evolvability. From this perspective, artificial ontogenies have largely been seen as a vehicle to relate the indirect encodings to complex phenotypes. However, the introduction of a development phase does not come without other consequences. We show that the conjunction of the latent ontogenic stucture and the common practice of only evaluating the final phenotype obtained from development can have a net retarding effect on evolution. Using a formal model of development, we show that this effect arises primarily due to the relation between the ontogenic structure to the fitness function, which in turn impacts the properties being evaluated and selected for during evolution. This effect is empirically demonstrated with a toy search problem using LOGO-turtle based embryogenic processes
Configuration Analysis to Support Product Redesign for End-of-life Disassembly
No full textEnd-of-life product disassembly is an important process that makes the parts of a product available for different material and part recycling processes at end of its useful life. However, the efficiency of the disassembly process greatly affects the economics of meeting the environmental goals set for the product. An important determinant of the efficiency of disassembly is the product configuration. Therefore, it is essential for the designer to assess these implications of the configuration while designing a product for end-of-life disassembly. In this paper, a formal model, called the Configuration-Value (CV) model, is proposed to evaluate and analyse the effect of configuration on disassembly. The model focuses on the rate of value extraction during the disassembly process. The model is used to identify the critical bottlenecks in the configuration, to help the designer to identify the design changes that need to be made to improve the product \u27disassemblability\u27. An example is presented to demonstrate the application of the proposed model
- …
