Ecological Neural Architecture Search
When employing an evolutionary algorithm to optimize a neural networks architecture, developers face the added challenge of tuning the evolutionary algorithm's own hyperparameters - population size, mutation rate, cloning rate, and number of generations. This paper introduces Neuvo Ecological Neural Architecture Search (ENAS), a novel method that incorporates these evolutionary parameters directly into the candidate solutions' phenotypes, allowing them to evolve dynamically alongside architecture specifications. Experimental results across four binary classification datasets demonstrate that ENAS not only eliminates manual tuning of evolutionary parameters but also outperforms competitor NAS methodologies in convergence speed (reducing computational time by 18.3%) and accuracy (improving classification performance in 3 out of 4 datasets). By enabling "greedy individuals" to optimize resource allocation based on fitness, ENAS provides an efficient, self-regulating approach to neural architecture search.
Code (0)
등록된 구현이 없습니다.
Tasks
Binary ClassificationNeural Architecture SearchMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
The Multilayer Nature of Ecological Networks
Although networks provide a powerful approach to study a large variety of ecological systems, their formulation does not typically account for multiple interaction types, interactions that vary in space and time, and int…
Deep learning-based ecological analysis of camera trap images is impacted by training data quality and quantity
Large image collections generated from camera traps offer valuable insights into species richness, occupancy, and activity patterns, significantly aiding biodiversity monitoring. However, the manual processing of these d…
Deep LearningComparing cars with apples? Identifying the appropriate benchmark countries for relative ecological pollution rankings and international learning
Research in Data Envelopment Analysis has created rankings of the ecological efficiency of countries' economies. At the same time, research in economic complexity has provided new methods to depict productive structures …
On the stability of competitive ecological communities with pairwise and higher-order interactions
Ecological communities intrigue researchers seeking to explain the emergence of biodiversity observed in nature. This raises a fundamental question. What sustains the stability and coexistence of species within these eco…
Towards Ecologically Valid Research on Language User Interfaces
Language User Interfaces (LUIs) could improve human-machine interaction for a wide variety of tasks, such as playing music, getting insights from databases, or instructing domestic robots. In contrast to traditional hand…
BIG-bench Machine Learningvalid