Publication
A first approach to some basilar questions on the relationship between state-of-the-art GP variants and GEP
| datacite.subject.fos | Ciências Naturais::Ciências da Computação e da Informação | pt_PT |
| dc.contributor.advisor | Oliveira, José Valente de | |
| dc.contributor.author | Neto, David Benedy Pereira | |
| dc.date.accessioned | 2019-08-01T10:01:36Z | |
| dc.date.available | 2019-08-01T10:01:36Z | |
| dc.date.issued | 2019-01-30 | |
| dc.date.submitted | 2018 | |
| dc.description.abstract | Gene Expression Programming (GEP) is an alternative to Genetic Programming (GP). Given its characteristics compared to GP, we question if GEP should be the standard choice for evolutionary program synthesis, both as base for research and practical application. We raise the question if such a shift could increase the rate of investigation, applicability and the quality of results obtained from evolutionary techniques for code optimization. We present three distinct and unprecedented studies using GEP in an attempt to develop understanding, investigate the potential and forward the branch. Each study has an individual contribution on its own involving GEP. As a whole, the three studies try to investigate di erent aspects that might be critical to answer the questions raised in the previous paragraph. In the rst individual contribution, we investigate GEP's applicability to automatically synthesize sorting algorithms. Performance is compared against GP under similar experimental conditions. GEP is shown to be capable of producing sorting algorithms and outperforms GP in doing so. As a second experiment, we enhanced GEP's evolutionary process with semantic awareness of candidate programs, originating Semantic Gene Expression Programming (SGEP), similarly to how Semantic Genetic Programming (SGP) builds over GP. Geometric semantic concepts are then introduced to SGEP, forming Geometric Semantic Gene Expression Programming (GSGEP). A comparative experiment between GP, GEP, SGP and SGEP is performed using di erent problems and setup combinations. Results were mixed when comparing SGEP and SGP, suggesting performance is signi cantly related to the problem addressed. By out-performing the alternatives in many of the benchmarks, SGEP demonstrates practical potential. The results are analyzed in di erent perspectives, also providing insight on the potential of di erent crossover variations when applied along GP/GEP. GEP' compatibility with innovation developed to work with GP is demonstrated possible without extensive adaptation. Considerations for integration of SGEP are discussed. In the last contribution, a new semantic operator is proposed, SCC, which applies crossover conditionally only when elements are semantically di erent enough, performing mutation otherwise. The strategy attempts to encourage semantic diversity and wider the portion of the semantic-solution space searched. A practical experiment was performed alternating the integration of SCC in the evolutionary process. When using the operator, the quality of obtained solutions alternated between slight improvements and declines. The results don't show a relevant indication of possible advantage from its employment and don't con rm what was expected in the theory. We discuss ways in which further work might investigate this concept and assess if it has practical potential under di erent circumstances. On the other hand, in regards to the basilar questions of this investigation, the process of development and testing of SCC is performed completely on a GEP/SGEP base, suggesting how the latest can be used as the base for future research on evolutionary program synthesis. | pt_PT |
| dc.description.abstract | Programa c~ao Gen etica por Express~oes (GEP) e uma alternativa recente a Programa c~ao Gen etica (GP). Neste estudo observamos o GEP e colocamos a quest~ao se este n~ao deveria ser tratado como primeira escolha quando se trata de sintetiza c~ao autom atica de programas atrav es de m etodos evolutivos. Dadas as caracteristicas do GEP perguntamonos se esta mudan ca de perspectiva poderia aumentar a investiga c~ao, aplicabilidade e qualidade dos resultados obtidos para a optimiza c~ao de c odigo por m etodos evolutivos. Neste estudo apresentamos tr^es contribui c~oes in editas e distintas usando o algoritmo GEP. Cada uma das contribui c~oes apresenta um avan co ou investiga c~ao no campo da GEP. Como um todo, estas contribui c~oes tentam obter cohecimento e informa c~oes para se abordar a quest~ao geral apresentada no p aragrafo anterior. Na primeira contribui c~ao, investiga-mos e testamos o GEP no problema da sintese autom atica de algoritmos de ordena c~ao. Para o melhor do nosso conhecimento, esta e a primeira vez que este problema e abordado com o GEP. A performance e comparada a do GP em condi c~oes semelhantes, de modo a isolar as caracteristicas de cada algoritmo como factor de distin c~ao. As a second experiment, we enhanced GEP's evolutionary process with semantic awareness of candidate programs, originating Semantic Gene Expression Programming (SGEP), similarly to how Semantic Genetic Programming (SGP) builds over GP. Geometric semantic concepts are then introduced to SGEP, forming Geometric Semantic Gene Expression Programming (GSGEP). A comparative experiment between GP, GEP, SGP and SGEP is performed using di erent problems and setup combinations. Results were mixed when comparing SGEP and SGP, suggesting performance is signi cantly related to the problem addressed. By out-performing the alternatives in many of the benchmarks, SGEP demonstrates practical potential. The results are analyzed in di erent perspectives, also providing insight on the potential of di erent crossover variations when applied along GP/GEP. GEP's compatibility with innovation developed to work with GP is demonstrated possible without extensive adaptation. Considerations for integration of SGEP are discussed. Na segunda contribui c~ao, adicionamos ao processo evolutivo do GEP a capacidade de medir o valor sem^antico dos programas que constituem a popula c~ao. A esta variante damos o nome de Programa c~ao Gen etica por Express~oes Sem^antica (SGEP). Esta variante tr as para o GEP as mesmas caracteristicas que a Programa c~ao Gen etica Sem^antica(SGP) trouxe para o GP convencional. Conceitos geom etricos s~ao tamb em apresentados para o SGEP, extendendo assim a variante e criando a Programa c~ao Gen etica por Express~oes Geom etrica Sem^antica (GSGEP). De forma a testar estas novas variantes, efectuamos uma experi^encia onde s~ao comparados o GP, GEP, SGP e SGEP entre diferentes problemas e combina c~oes de operadores de cruzamento. Os resultados mostraram que n~ao houve um algoritmo que se destaca-se em todas as experi^encias, sugerindo que a performance est a signi cativamente relacionada com o problema a ser abordado. De qualquer modo, o SGEP obteve vantagem em bastantes dos benchmarks, dando assim ind cios de pot^encial ter utilidade pr atica. De um modo geral, esta contribui c~ao demonstra que e possivel utilizar tecnologia desenvolvida a pensar em GP no GEP sem grande esfor co na adapta c~ao. No m da contribui c~ao, s~ao discutidas algumas considera c~oes sobre o SGEP. Na terceira contribui c~ao propomos um novo operador, o Cruzamento Sem^antico Condicional (SCC). Este operador, baseado na dist^ancia sem^antica entre dois elementos propostos, decide se os elementos s~ao propostos para cruzamento, ou se um deles e mutato e ambos re-introduzidos na popula c~ao. Esta estrat egia tem como objectivo aumentar a diversidade gen etica na popula c~ao em fases cruciais do processo evolutivo e alargar a por c~ao do espa co sem^antico pesquisado. Para avaliar o pot^encial deste operador, realizamos uma experi^encia pr atica e comparamos processos evolutivos semelhantes onde o uso ou n~ao uso do SCC e o factor de distin c~ao. Os resultados obtidos n~ao demonstraram vantagens no uso do SCC e n~ao con rmam o esperado em teoria. No entanto s~ao discutidas maneiras em que o conceito pode ser reaproveitado para novos testes em que possa ter pot^encial para demonstrar resultados possitivos. Em rela c~ao a quest~ao central da tese, visto este estudo ter sido desenvolvido com base em GEP/SGEP e visto a teoria do SCC ser compativel com GP, e demonstrado que um estudo geral a area da sintese de algoritmos por meios evolutivos, pode ser conduzido com base no GEP. | pt_PT |
| dc.identifier.tid | 202230716 | pt_PT |
| dc.identifier.uri | http://hdl.handle.net/10400.1/12710 | |
| dc.language.iso | eng | pt_PT |
| dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | pt_PT |
| dc.subject | First approach | pt_PT |
| dc.subject | Basilar questions | pt_PT |
| dc.subject | Relationship between State-of-the Art | pt_PT |
| dc.subject | GP variants and GEP | pt_PT |
| dc.title | A first approach to some basilar questions on the relationship between state-of-the-art GP variants and GEP | pt_PT |
| dc.title.alternative | Automatic synthesis of sorting algorithms by gene expression programming + (geometric) semantic gene expression programming + encouraging phenotype variation with a new semantic operator: semantic conditional crossover | pt_PT |
| dc.type | master thesis | |
| dspace.entity.type | Publication | |
| rcaap.rights | openAccess | pt_PT |
| rcaap.type | masterThesis | pt_PT |
| thesis.degree.grantor | Universidade do Algarve. Faculdade de Ciências e Tecnologia | |
| thesis.degree.level | Mestre | |
| thesis.degree.name | Mestrado em Engenharia Informática | pt_PT |