首页|Research Conducted at Victoria University Wellington Has Provided New Informatio n about Machine Learning (Modular Multitree Genetic Programming for Evolutionary Feature Construction for Regression)
Research Conducted at Victoria University Wellington Has Provided New Informatio n about Machine Learning (Modular Multitree Genetic Programming for Evolutionary Feature Construction for Regression)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-Investigators publish new report on Ma chine Learning. According to news reporting originating in Wellington, New Zeala nd, by NewsRx journalists, research stated, "Evolutionary feature construction i s a key technique in evolutionary machine learning, with the aim of constructing high-level features that enhance performance of a learning algorithm. In real-w orld applications, engineers typically construct complex features based on a com bination of basic features, reusing those features as modules." Funders for this research include Marsden Fund (NZ), Science for Technological I nnovation Challenge (SfTI) Fund, New Zealand Ministry of Business, Innovation an d Employment (MBIE), Huayin Medical, New Zealand Ministry of Business, Innovatio n and Employment (MBIE). The news reporters obtained a quote from the research from Victoria University W ellington, "However, modularity in evolutionary feature construction is still an open research topic. This article tries to fill that gap by proposing a modular and hierarchical multitree genetic programming (GP) algorithm that allows trees to use the output values of other trees, thereby representing expressive featur es in a compact form. Based on this new representation, we propose a macro paren t-repair strategy to reduce redundant and irrelevant features, a macro crossover operator to preserve interactive features, and an adaptive control strategy for crossover and mutation rates to dynamically balance the tradeoff between explor ation and exploitation. A comparison with seven bloat control methods on 98 regr ession datasets shows that the proposed modular representation achieves signific antly better results in terms of test performance and smaller model size."
WellingtonNew ZealandAustralia and N ew ZealandCyborgsEmerging TechnologiesGenetic ProgrammingGeneticsMachi ne LearningVictoria University Wellington
NETL
NSTL
